THE DEVELOPMENT OF GIPSICAM v3 A MOBILE MAPPING SYSTEM FOR RAPID ROAD ASSET DATA CAPTURE
By
Dennis Robert Entriken
B.Sc., James Cook University, Australia, 1992
A thesis submitted to The University of New South Wales in partial fulfilment of the requirements for the degree of Master of Philosophy
School of Surveying and Spatial Information Systems The University of New South Wales Sydney, NSW 2053, Australia
March 2011
2
ORIGINALITY STATEMENT
'I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person. or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution. except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work. except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.'
Signed ... 1).~ .....~ ......
Date ...~ . !/ .J./.;J.. .C>. !l ......
3
COPYRIGHT STATEMENT
'I hereby grant the University of New South Wales or ~s agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation. I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstract International (this is applicable to doctoral theses only). I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.'
Signed ... (/) .~...... ~ ......
Date .. .3.1 I .3. /.;;,.o.. !.t......
AUTHENTICITY STATEMENT
"I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.'
Signed .. ..'i) ~ .....~ ......
Date ...... ?. .t/. .3/..?P..II......
4 Abstract
ABSTRACT
The use of a Mobile Mapping System (MMS) to quickly collect large amounts of data has been proven to be a cost effective and very beneficial technology for Road Authorities around the world. Early in the development of MMS technology the Roads and Traffic Authority of NSW (RTA) recognised the benefits of utilising such a system in the management and maintenance of their state road network and developed the first generation GIPSICAM system. GIPSICAM, which is an acronym for “Global and Inertial Positioning Systems with Image Capture for Asset Management”, is the RTA’s in-house developed, operated and maintained vehicle-mounted MMS.
The first generation GIPSICAM system (GCv1) had no stable configuration and was modified for each project it was used for. Then in 1999 the second generation GIPSICAM system (GCv2) was developed for the lead up to the 2000 Olympic Games. Development of GCv2 continued after the Olympic Games and its operation was institutionalised in the annual surveys of NSW state roads from 2001 onwards.
In July 2004 a strategic report on RTA Pavement Condition Monitoring Equipment was released that identified GIPSICAM data as an important RTA corporate dataset. However, the report also identified serious reliability issues with GCv2 due to the ad- hoc nature of its development and the age of some of its technology and equipment. In early 2005 the decision was made to develop the third generation GIPSICAM system (GCv3), also known as “GIPSICAM Version 3”.
The objective of the development work was to take advantage of the experience from previous generation GIPSICAM systems in order to design a MMS that utilises current technology to ensure that the RTA has a reliable, safe, efficient, accurate, high quality GIPSICAM capability for the purpose of rapid road asset data capture and other applications of MMS technology.
GCv3 was completed in mid-2006 and officially commenced operations in September 2006, at the start of the 2006/2007 GIPSICAM survey season.
i Abstract
This thesis outlines the development of GCv3, from the selection and modification of the vehicle, the selection and integration of component technologies, and the development of operational procedures for rapid road asset data capture.
ii Acknowledgments
ACKNOWLEDGMENTS
This research was undertaken on a part-time basis during the period from February 2005 through to September 2006 under the supervision of Professor Chris Rizos, from the University of NSW, and Mr Steve Greening, from the Roads and Traffic Authority of NSW. I am sincerely grateful to them both for their encouragement and guidance throughout this project.
I wish to thank my colleagues at the NSW Roads and Traffic Authority, for their support and assistance. In particular I wish to thank Mr Michael Vernon for sharing his vast experience as an operator in the previous GIPSICAM vehicle out-in-the-field and his handyman / tradesman skills that came in very useful in the planning and fit-out of the new vehicle. I also wish to thank Mr Ken Root and Mr Brendan Root for the programming work they undertook modernising the GIPSICAM software and incorporating the new features and functionality of GCv3 into the GIPSICAM software suite. Finally, again I wish to thank Mr Steve Greening for taking me under-his-wing as an understudy of the work he had undertaken on previous generations of the GIPSICAM system as part of the RTA’s knowledge transfer process before he retired, for allowing me the opportunity to develop the next generation GIPSICAM system, and for his coaching and mentoring to enable me to take on the role of managing the GIPSICAM system after he retired.
I wish to thank my RTA colleagues (some of who have now retired) that supported the development of the next generation GIPSICAM system. In particular I wish to thank Mr Ron Ferguson and Mr David Pratt, who provided funding for the development of GCv3. I also wish to thank my senior management colleagues, Mr Peter Collins, Mr Michael Bushby, Mr Chris Harrison, Mr Steve Dunlop and Mr Mark Gordon, for their far- sighted and continued support of the GIPSICAM system.
I wish to thank David Morphett and the guys from ETT, Brett Franzi from Industrial Evolution, Kevin Dowsey from Total-turkey solutions, and Ron Elliott from ARRB.
iii Acknowledgments
I wish to thank my wife and best friend Lihua for her patience and encouragement, particularly during the time I was writing up this thesis, when I was struggling for time and motivation to complete the write-up while working full time under the pressures of a never-ending business review, the constant possibility of the outsourcing of our work, the stress of managing a team of people who are trying to cope with the ongoing uncertainty and change in the workplace, and the limitations in terms of permanent resources. Lihua also provided assistance in proofreading this thesis, which is very much appreciated.
Last, but not least, I wish to thank my parents Neil and Robena Entriken for providing me with the means to acquire an education and giving me the best chance they could to “live long and prosper” in this modern world.
Views expressed in this thesis are those of the author, and are not necessarily the views of the Roads and Traffic Authority of NSW.
iv Table of Contents
TABLE OF CONTENTS
ABSTRACT ...... i ACKNOWLEDGMENTS ...... iii TABLE OF CONTENTS ...... v LIST OF FIGURES ...... x LIST OF TABLES ...... xiv GLOSSARY AND DEFINITIONS ...... xv ACRONYMS AND INITIALISMS ...... xvi CHAPTER 1 ...... 1 INTRODUCTION...... 1 1.1 Introduction ...... 1 1.2 What is a Mobile Mapping System (MMS)? ...... 1 1.3 What is GIPSICAM? ...... 3 1.4 A brief history and overview of GIPSICAM v1 (GCv1) ...... 3 1.5 A brief history and overview of GIPSICAM v2 (GCv2) ...... 5 1.6 Why the need to upgrade GIPSICAM? ...... 7 1.7 Along came Dennis Entriken ...... 11 1.8 A brief overview of GIPSICAM v3 (GCv3) ...... 13 1.9 Motivation ...... 18 1.10 Research objectives ...... 18 1.11 Contribution of the research ...... 19 1.12 Outline of the thesis ...... 19 CHAPTER 2 ...... 21 REVIEW OF MMS TECHNOLOGY ...... 21 2.1 Introduction ...... 21 2.2 Trial outsourcing of road image collection in FY05/06 ...... 23 2.2.1 Background ...... 23 2.2.2 Roads requested to be surveyed during FY05/06 ...... 24 2.2.3 Roads surveyed by ARRB during FY05/06 ...... 25 2.2.4 Visual assessment of road video image quality ...... 25 2.2.5 Reasons for classifying road video image as unacceptable ...... 25
v Table of Contents
2.2.6 Coverage of roads requested in survey plan ...... 27 2.3 An overview of positioning and attitude technology utilised in GCv3 ...... 28 2.3.1 Global Positioning System (GPS) ...... 28 2.3.2 Inertial Navigation System (INS) ...... 29 2.3.3 Odometer ...... 31 2.4 Reference Systems used in the RTA ...... 33 2.4.1 RTA Lambert94 ...... 34 2.4.2 Map Grid of Australia (MGA) ...... 34 2.4.3 RTA RoadLoc ...... 35 CHAPTER 3 ...... 37 VEHICLE SELECTION AND MODIFICATION ...... 37 3.1 Introduction ...... 37 3.2 Vehicle selection ...... 37 3.3 Vehicle modification ...... 42 3.3.1 Equipment/storage cabinet ...... 44 3.3.2 Cargo barrier ...... 49 3.3.3 Air conditioning ...... 50 3.3.4 Cargo area insulating and soundproofing ...... 52 3.3.5 Equipment electrical system ...... 53 3.3.6 Camera turret ...... 65 3.3.7 Computer/equipment rack ...... 74 3.3.8 Equipment operator work area ...... 76 3.3.9 GIPSITRAC odometer ...... 79 3.3.10 GPS antenna mount points ...... 80 3.3.11 Survey/warning lights ...... 83 3.3.12 Magnetic signage, stickers and RTA decals ...... 83 3.3.13 General/minor modifications ...... 85 3.4 Concluding remarks ...... 86 CHAPTER 4 ...... 90 EQUIPMENT SELECTION AND SYSTEMS INTEGRATION ...... 90 4.1 Introduction ...... 90 4.2 Equipment selection ...... 90 4.2.1 DGPS receiver with real-time differential corrections ...... 90
vi Table of Contents
4.2.2 GPS receiver for post-processed differential corrections ...... 93 4.2.3 ARRB GIPSITRAC ...... 94 4.2.4 FOG Gyroscope ...... 99 4.2.5 Vehicle odometers ...... 100 4.2.6 Video equipment ...... 101 4.2.7 GCv3 onboard computers and data loggers ...... 109 4.2.8 Data storage ...... 111 4.3 Equipment integration ...... 112 4.3.1 Primary absolute position determination subsystem ...... 112 4.3.2 Backup absolute position determination subsystem ...... 114 4.3.3 GCAM subsystem ...... 116 4.3.4 Integration of the subsystems ...... 128 4.4 Concluding remarks ...... 128 CHAPTER 5 ...... 131 SOFTWARE ...... 131 5.1 Introduction ...... 131 5.1 GCv3 vehicle software ...... 131 5.1.1 GIPSICAM GCAM ...... 131 5.2 GCv3 processing software ...... 133 5.2.1 GIPSICAM GCAPTURE ...... 133 5.3 GCv3 asset capture software ...... 135 5.3.1 GIPSICAM AssetLoc ...... 135 5.4 GCv3 dataset display software ...... 136 5.4.1 GIPSICAM RoadFlix ...... 136 5.4.2 GIPSICAM Road Geometry Analyst ...... 137 5.4.3 GIPSICAM RoadBrowser ...... 138 5.4.4 ESRI ArcGIS ...... 139 CHAPTER 6 ...... 140 OPERATIONAL PROCEDURES ...... 140 6.1 Introduction ...... 140 6.2 Calibration procedures ...... 140 6.2.1 Camera/Lens calibration ...... 140 6.2.2 GIPSITRAC bench calibration ...... 142
vii Table of Contents
6.2.3 GIPSITRAC controlled 180 calibration ...... 143 6.2.4 GIPSITRAC suspension/tilt calibration ...... 145 6.2.5 GIPSITRAC odometer calibration ...... 147 6.2.6 GIPSITRAC daily 180 calibration ...... 149 6.2.7 GIPSITRAC daily 360 calibration ...... 150 6.3 Validation procedures ...... 151 6.3.1 GPS position validation ...... 151 6.3.2 GPS + inertial position validation ...... 152 6.3.3 Camera/lens calibration validation ...... 153 6.3.4 Road geometry validation ...... 157 6.3.5 Odometer distance validation ...... 158 6.3.6 GIPSITRAC controlled grade and crossfall validation ...... 158 6.4 Routine procedures...... 159 6.4.1 GCv3 GIPSITRAC start-of-season calibration and validation ...... 159 6.4.2 GCv3 vehicle daily hours of operation ...... 161 6.4.3 GCv3 GIPSICAM survey ...... 162 6.4.4 GCv3 GIPSICAM processing ...... 164 CHAPTER 7 ...... 166 TESTING AND VALIDATION...... 166 7.1 Introduction ...... 166 7.2 GCv3 system vehicle position and road alignment ...... 166 7.2.1 Display GCv3 vehicle positions in GIS software ...... 166 7.2.2 Comparison of GCv3 vehicle positions with Bangor Bypass survey data ...... 168 7.3 GCv3 system georeferenced road images ...... 169 7.3.1 GCv3 Rockdale test area ...... 169 7.3.2 Comparison with other data sources ...... 171 7.4 GCv3 system georeferenced road geometry ...... 172 7.4.1 Comparison of GCv3 road geometry data with Bangor Bypass survey data ...... 172 7.5 GCv3 system road image quality ...... 173 7.5.1 Optimal road image quality ...... 174 7.5.2 Hardware related road image quality problems ...... 174
viii Table of Contents
7.5.3 Configuration-related road image quality problems ...... 175 7.5.4 Procedural related road image quality problems ...... 178 7.6 GCv3 GIPSITRAC calibration and validation ...... 179 7.6.1 GCv3 GIPSITRAC start-of-season calibration and validation ...... 179 7.6.2 GIPSITRAC daily 180 calibration and daily 360 calibration ...... 181 7.6.3 Analysis of GIPSITRAC gyroscope and accelerometer calibration values ...... 183 CHAPTER 8 ...... 186 APPLICATION OF MMS TECHNOLOGY ...... 186 8.1 Introduction ...... 186 8.2 Routine applications of the GIPSICAM data within the RTA ...... 186 8.2.1 Road centreline vector data ...... 187 8.2.2 Georeferenced road images ...... 189 8.2.3 Georeferenced road geometry data ...... 191 8.3 Specific applications of the GIPSICAM data within the RTA ...... 195 8.3.1 Asset data collection ...... 195 8.3.2 Road pavement crack mapping ...... 196 8.3.3 Accurate measuring of distance ...... 196 8.3.4 School bus routes ...... 196 8.3.5 Road longsections (heights) ...... 197 8.3.6 Regional Forest Agreements project ...... 197 8.3.7 Historical record searches ...... 198 8.3.8 SCRIM site categories determination ...... 199 8.3.9 Point to Point Speed Zones (P2PZ) ...... 200 8.3.10 Speed Limiter Enforcement Zones (SLEZ) ...... 201 8.4 Concluding remarks ...... 202 CHAPTER 9 ...... 203 SUMMARY AND RECOMMENDATIONS ...... 203 9.1 The development of GIPSICAM v3 ...... 203 9.2 Future research and development opportunities for GIPSICAM ...... 204 9.3 Future of MMS technologies within a Roads and Traffic Authority ...... 206 9.4 Future of MMS technologies within commercial organisations ...... 207 REFERENCES ...... 209
ix List of Figures
LIST OF FIGURES
Figure 1. Example MMS vehicle positioning and acquisition technologies (ARRB, 2010) ...... 2 Figure 2. GCv2 vehicle at Rockdale Works Office ...... 5 Figure 3. The GCv2 operator work area ...... 6 Figure 4. The GCv2 digital video camera mounted on the roof of the vehicle...... 8 Figure 5. The GCv2 DGPS and GPS antennae mounted on the roof of the vehicle ...... 9 Figure 6. The GCv2 computer/equipment rack in the cargo area of the vehicle ...... 10 Figure 7. Mike Vernon, Steve Greening and Dennis Entriken ...... 11 Figure 8. The Roads and Traffic Authority’s GCv3 vehicle ...... 13 Figure 9. Overview of GCv3 sensor and equipment layout ...... 15 Figure 10. GCv3 vehicle sensor block diagram ...... 15 Figure 11. Front and side camera road images from GCv3 ...... 16 Figure 12. GCv3 road image positions displayed on ©SKM orthophotography data ... 17 Figure 13. ARRB Hawkeye 2000 vehicle ...... 22 Figure 14. Basic principle of a vibrating gyroscope (Bonsen et al, 2005)...... 31 Figure 15. Sensor signal generation in MR-sensors (Schmeißer et al, 1999) ...... 32 Figure 16. Front/Side view of KMI sensor and steel cog (Schmeißer et al, 1999 and modified) ...... 32 Figure 17. RTA Lambert94 parameters ...... 33 Figure 18. RoadLoc reference: 0000001,0010,B2,0.100,P ...... 36 Figure 19. Mercedes Sprinter 316 CDI specification (captured from Mercedes-Benz, 2003) ...... 41 Figure 20. The new Mercedes Sprinter 316 CDI ...... 42 Figure 21. More pictures of the new Mercedes Sprinter 316 CDI ...... 43 Figure 22. Rear view of cabinet showing disassembled storage area ...... 45 Figure 23. Rear view of cabinet showing five storage compartments ...... 46 Figure 24. Rear and top view of cabinet showing diesel “Jerry Can” ...... 46 Figure 25. Rear view of cabinet showing drawers ...... 47 Figure 26. Front and top view of structural walls of the cabinet ...... 48 Figure 27. Front view of the cabinet showing the seven storage areas (from L to R) ... 49
x List of Figures
Figure 28. The cargo barrier being manufactured (L) and installed (R) ...... 50 Figure 29. The 2nd air conditioner being installed (L) and when finished (R) ...... 51 Figure 30. Installation of the cargo area insulating and soundproofing ...... 53 Figure 31. The two Lifeline GPL-31T batteries were wired in parallel to provide 210Ah ...... 57 Figure 32. The engine bay being disassembled (L) and the Bosch alternator (R) ...... 58 Figure 33. The Ample Power Next Step Regulator NEXT-12P ...... 59 Figure 34. The distribution switch (L) and its installation on the passengers seat (R) .. 60 Figure 35. The interVOLT power conditioner ...... 61 Figure 36. The Sinergex power inverter ...... 62 Figure 37. External power connector (L) and Clipsal MDRC Housing (R) ...... 63 Figure 38. 12V power supply (L) and 5V power supply (R) ...... 64 Figure 39. GCv3 standard camera angles (horizontal plane relative to direction of travel) ...... 68 Figure 40. GCv3 standard camera angles (vertical plane relative to direction of travel) ...... 69 Figure 41. The initial design for the camera turret...... 70 Figure 42. The rectangular barrel concept ...... 71 Figure 43. Testing the new turret design ...... 73 Figure 44. The finished designs for the camera turret ...... 74 Figure 45. The computer/equipment rack ...... 75 Figure 46. The monitor arm and Dell monitor ...... 77 Figure 47. The equipment operator work area (the passenger side) ...... 78 Figure 48. The steel cog and KMI 15/1 rotational speed sensor ...... 80 Figure 49. GPS antenna mount points ...... 82 Figure 50. GCv3 with magnetic signage (front view) ...... 84 Figure 51. GCv3 with magnetic signage (rear view) ...... 85 Figure 52 Trimble Pro XRS, rear view ...... 91 Figure 53 Trimble Pro XRS specification (Trimble documentation, modified) ...... 92 Figure 54 Trimble Pro XL specification (Trimble documentation, modified) ...... 94 Figure 55. ARRB GIPSITRAC INS (with the top removed for maintenance) ...... 96 Figure 56 Motorola PVT6 specification (Motorola documentation, modified) ...... 97 Figure 57 Murata ENV specifications (Avnet Kopp, 2003) ...... 97
xi List of Figures
Figure 58 Honeywell QA700 specifications (Honeywell, 2004 and modified)...... 98 Figure 59. KVH DSP-3000 FOG ...... 100 Figure 60. Sony DFW-SX910 camera and Fujinon lens ...... 103 Figure 61. GCv3 standard camera HFOV (horizontal plane relative to direction of travel) ...... 108 Figure 62 Primary position determination system ...... 113 Figure 63 Community Base Station hardware setup (Trimble, 1997) ...... 115 Figure 64. GCAM main menu ...... 131 Figure 65. GCAM Acquire module showing a survey in progress ...... 132 Figure 66. GCAM Calibration module ...... 132 Figure 67. GCAPTURE showing unadjusted DGPS/INS data (RTA, 2011 GCAP)... 133 Figure 68. GCAPTURE vertical adjustment (RTA, 2011 GCAP) ...... 134 Figure 69. GCAPTURE video processing (RTA, 2011 GCAP) ...... 134 Figure 70. AssetLoc being used to measure the width of the lane at the Rockdale test area ...... 135 Figure 71. GIPSICAM road image annotation ...... 135 Figure 72. RoadFlix showing the GIPSICAM road images ...... 136 Figure 73. RGA showing GIPSICAM road geometry data ...... 137 Figure 74. RoadBrowser showing the location of test road MR667 ...... 138 Figure 75. GIPSICAM image points displayed in ESRI ArcGIS on SKM orthophoto ...... 139 Figure 76. Transverse suspension/tilt test ...... 146 Figure 77. Longitudinal suspension/tilt test ...... 147 Figure 78. GCv3 GIPSITRAC start of season calibration field sheet...... 160 Figure 79. GCv3 GIPSICAM survey run-sheet ...... 163 Figure 80. GCv3 GIPSICAM processing check-sheet ...... 165 Figure 81. Comparison of GCv3 vehicle positions with 10cm orthophotography ...... 167 Figure 82. Example GCv3 high resolution road image ...... 174 Figure 83. An example of vertical smear ...... 175 Figure 84. An example of the “smoky” effect ...... 176 Figure 85. An example of sunlight shining directly on the camera lens ...... 177 Figure 86. An example of the mirror effect ...... 178 Figure 87. GCv3 GIPSITRAC start of season calibration results ...... 180
xii List of Figures
Figure 88. GCv3 georeferenced road image ...... 190 Figure 89. A sample plot of GCv3 road geometry data ...... 194 Figure 90. Sample GIPSICAM centreline data used in the RFA project ...... 198
xiii List of Tables
LIST OF TABLES
Table 1. Road video data collection survey plan (FY05/06) provided to ARRB ...... 24 Table 2. Summary of road video image visual quality ...... 25 Table 3. Rockdale coordinated ground stations, referenced in MGA zone 56 ...... 152 Table 4. GCv3 Rockdale test area detailed survey data ...... 156 Table 5. Guide to GCv3 hours of operation in metropolitan areas ...... 161 Table 6. Guide to GCv3 hours of operation in non-metropolitan areas ...... 161 Table 7. Difference of single observation (AHD/RTA Lambert94) from survey data 168 Table 8. Lane width validation at GCv3 Rockdale test area ...... 170 Table 9. Segment length validation at GCv3 Rockdale test area ...... 170 Table 10. GCv3 lane widths (Nov 2008) vs orthophotography lane widths ...... 171 Table 11. Difference of single observation (grade/crossfall as %) from survey data .. 172 Table 12. GIPSITRAC gyro and accelerometer calibration values over two weeks ... 184
xiv Glossary and Definitions
GLOSSARY AND DEFINITIONS
Mega-Pixel A term referring to digital camera technology which supports a resolution of greater than or equal to 1M pixels.
xv Acronyms and Initialisms
ACRONYMS AND INITIALISMS
ABS Anti-lock Braking System AGM Absorbed Glass Mat Ah Amp Hours AHD Australian Height Datum ARAN Automatic Road Analyser ARRB Australian Road Research Board BI Business Intelligence CBD Central Business District CCA Cold Cranking Amps CCD Charged Coupled Device CCTV Closed Circuit Television CDI Common rail Direct fuel Injection CMOS Complementary Metal-Oxide Semiconductor CORS Continuously Operating Reference Stations DGPS Differential Global Positioning System DIN Deutsches Institut für Normung (German Institute for Standardization) DR Dead Reckoning DV Digital Video DVI-D Digital Visual Interface - Digital EDW Enterprise Data Warehouse emf Electromotive Force EMF Electromagnetic Field ETT Emergency Transport Technology Pty Ltd FOG Fibre Optic Gyroscope FOV Field of View GA Geoscience Australia GCv1 GIPSICAM System version 1 GCv2 GIPSICAM System version 2 GCv3 GIPSICAM System version 3 GDA94 Geocentric Datum of Australia 1994
xvi Acronyms and Initialisms
GIPSICAM Global and Inertial Positioning Systems Image Capture for Asset Management GIPSITRAC Global and Inertial Positioning Systems Integration for Tracing Route Alignment and Crossfall GIS Geographic Information System GLONASS Global Navigation Satellite System (Russian GNSS) GNSS Global Navigation Satellite System GPS Global Positioning System HDD Hard Disk Drive HDOP Horizontal Dilution of Precision HFOV Horizontal Field of View HP High Performance INS Inertial Navigation System LiDAR Light Detection And Ranging LPMA Land & Property Management Authority of NSW LTS Location Translation Service MCB Miniature Circuit Breaker MDRC Modular DIN Rail Component MEMS Micro-Electro Mechanical System MGA Map Grid of Australia MHIS Mobile Highway Inventory System MMS Mobile Mapping System MIT Massachusetts Institute of Technology OEM Original Equipment Manufacturer OH&S Occupational Health and Safety OR Olympic Route PC Personal Computer PFCBS Pathfinder Community Base Station PTR Principal Transport Route QMS Quality Management System RAID Redundant Array of Independent Disks RCBO Residual Circuit Breaker with Overload protection RGDAS Road Geometry Data Acquisition System
xvii Acronyms and Initialisms
RMIT Royal Melbourne Institute of Technology RPM Revolutions per Minute RTA Roads and Traffic Authority of NSW RTK Real-Time Kinematic R&D Research and Development SA Selective Availability SBAS Satellite Based Augmentation System SCRIM Sideways Force Coefficient Routine Investigation Machine SMH Sydney Morning Herald SSM State Survey Mark TMC Transport Management Centre UNSW University of New South Wales UoM University of Melbourne UPS Uninterruptible Power Supply USB Universal Serial Bus USNO US Naval Observatory UTM Universal Transverse Mercator V Volts VBS Virtual Base Station VGA Video Graphics Array VFOV Vertical Field of View VISAT Video Inertial Satellite WGS84 World Geodetic System 1984
xviii Introduction
CHAPTER 1 INTRODUCTION
1.1 Introduction
The primary objective of a Roads and Traffic Authority is the management of the road network within its jurisdiction, to ensure efficient traffic flow and safe roads. The road network may comprise of hundreds or even thousands of kilometres of road carriageway, consisting of various asset components such as road pavements, lanes, road shoulders, bridges, tunnels, culverts, vehicle ferries, rest areas, traffic lights, roundabouts, medians, guardrails, guideposts, signs and line marking (Roads and Traffic Authority, 2007).
An important part of managing a road network is building and maintaining an inventory of assets. Xiong and Floyd (2004) stated that a “Vehicle-based Mobile Mapping System (MMS) proved to be an effective technology for sign inventory and has the potential for many other types of roadway features and characteristics” and that “these features can be very effectively captured with MMS images.” This statement is verified by the experience of the Roads and Traffic Authority of NSW (RTA) with their GIPSICAM MMS. However, an MMS has much more to offer than just building an inventory of assets.
1.2 What is a Mobile Mapping System (MMS)?
The classical definition of a MMS is: a “kinematic platform, upon which multiple sensors have been integrated and synchronized to a common time base, to provide 3D near continuous and automatic positioning of both the platform and simultaneously collected geo-spatial data” (Grejner-Brzezinska, 2004).
However, the definition of a MMS may vary depending on the context of the solution. Hand held GPS manufacturers market their devices as a MMS, capable of collecting and
1 Introduction
editing GPS track data in the field. GIS software vendors market their PDA based applications as MMS enabling software capable of collecting and editing spatial data in the field. These solutions consist of a GPS-enabled hand held device with software to allow the capture and editing of spatially referenced data. The data capture and data editing are manual processes and usually involve moving from asset to asset, or point- of-interest to point-of-interest as a discrete process.
Aerial- and land-based vehicle MMS usually consist of one or more positioning technologies, such as GPS and INS, that are integrated with one or more data acquisition technologies, such as photography, video, laser scanners, LiDAR and ground penetrating radar (See Figure 1). They generally collect large amounts of data and do not edit collected data in the field. The data collection is typically an automated continuous process (fully automated or semi-automated) that ‘captures’ any assets that pass within the field of view of the mass data acquisition device(s). These types of MMS have been installed in aeroplanes, helicopters, cars, trucks, vans, trains, and most recently in remote-controlled aerial and land-based drones.
Figure 1. Example MMS vehicle positioning and acquisition technologies (ARRB, 2010)
2 Introduction
In the context of the RTA’s GIPSICAM vehicle (See Figure 2), the definition of a MMS matches the latter of the two MMS descriptions, consisting of a vehicle with multiple integrated sensors and mass data acquisition technologies to automatically and continuously capture a record of the road network and associated road side assets, and enable the extraction of road alignment, road geometry and road side asset information.
1.3 What is GIPSICAM?
GIPSICAM is a MMS, developed in-house by the RTA to permit the survey of the NSW state road network. The word GIPSICAM is an acronym for “Global-Inertial Positioning Systems Image Capture for Asset Management” (Greening, 2003).
1.4 A brief history and overview of GIPSICAM v1 (GCv1)
In the early 1990’s the RTA realised the potential of rapid data acquisition techniques to efficiently collect large volumes of asset data along the state roads of NSW. It was found that this type of data collection was more efficient than manual collection techniques in terms of cost and time required for survey. During this time the RTA experimented with simple Video Log systems as a precursor to the development of GIPSICAM. Archived data still exists from some of the early RTA MMS surveys, going back as far as April 1993.
In the early 1990’s a joint development effort between ARRB, the Australian Road Authorities, RMIT and the University of Melbourne led to the development of the Road Geometry Data Acquisition System (RGDAS) and GIPSITRAC. GIPSITRAC is an acronym for “Global and Inertial Positioning System Integration for Tracking Route Alignment and Crossfall” (ARRB, 1995). In its initial configuration GIPSITRAC consisted of a tightly coupled integration of gyroscopes, accelerometers, GPS and a sensor to measure distance, which was utilised to collect road geometry and road alignment data. Later a video time-code interface was integrated into GIPSITRAC to
3 Introduction
allow the synchronisation of video, as part of a joint development effort between ARRB and the RTA, which led to the development of the first generation GIPSICAM vehicle by the RTA in 1996-97. The first generation GIPSICAM system (GCv1) was the “first road video inventory system based on merged DGPS and inertial data” in Australia (Greening, 2003).
GCv1 did not have a stable configuration during its life. It was modified and adapted for each project it was used in. However the configuration was always based around CCTV cameras, analogue video, DGPS and the GIPSITRAC INS. GCv1 was utilised in 1996- 97 to collect road video data for the proposed upgrade of the Pacific Highway. During this project GCv1 utilised two cameras that were aligned for stereo positioning. In 1997 GCv1 was used to measure road cracking on the Sydney to Newcastle Freeway, with the cameras facing down at the road pavement. Then in 1998, using just one forward facing camera, GCv1 was used to capture road video data for the Tasmanian state road network. GCv1 was decommissioned in 1999, partly due to an accident and partly due to the upgrades required to provide a standardised system for the future.
4 Introduction
1.5 A brief history and overview of GIPSICAM v2 (GCv2)
In 1999, prior to the Sydney Olympics in 2000, the RTA’s Transport Management Centre (TMC) realised the potential benefits of having road video of the Olympic Routes (OR) and the Principal Transport Routes (PTR) for planning and incident response purposes, hence the TMC provided funding to survey these routes using GIPSICAM. This project led to the development of the second generation GIPSICAM system (GCv2), with the primary upgrades consisting of a change from analogue video to digital video to increase the quality of the road video data, and the further development of software for the processing and display of the data. The OR and the PTR were surveyed using GCv2 in 1999 and 2000. GIPSICAM v2 (GCv2) was developed in-house by the RTA in 1999 (See Figure 2).
Figure 2. GCv2 vehicle at Rockdale Works Office
5 Introduction
Figure 3. The GCv2 operator work area
The core components of the GCv2 vehicle were: Volkswagen Transporter T4, 2.5L inline 5 cylinder, 4-door minivan vehicle, with after-market modifications (See Figure 2). DGPS receiver with real-time differential corrections, which provided the primary absolute positioning of the vehicle trajectory (See Figure 5). GPS receiver, which was the backup absolute positioning instrument, logged autonomous GPS data that could be post-processed with GPS base station data to provide supplementary absolute position data, if or when required. Digital readout odometers, which used a sensor on the rear RHS wheel to determine distance travelled. Sony Handycam DCR TRV900 (See Figure 4) and JVC mini DV recorder (See Figure 6), which were used to collect forward facing, PAL resolution, digital video. Operator console consisting of a laptop computer, mounted on a bracket that was fixed to the dash on the passenger side of the vehicle (See Figure 3).
6 Introduction
Four lead-acid 12V batteries and a 240V power inverter that provided power to the vehicle equipment. GIPSITRAC INS, providing relative positioning and vehicle attitude, comprising of: o Video time-code interface, which provided a mechanism for synchronisation of the digital video with the other vehicle sensors. o GPS receiver, which was used for GIPSITRAC sensor synchronisation. o Two micro-electromechanical system (MEMS) gyroscopes, which were used to determine the direction of travel and the horizontal radius of the vehicle path. o Two accelerometers, one placed in a longitudinal orientation and the other in a transverse orientation, which were used to determine the grade, cross-fall and vertical radius of the vehicle path. o Rotational speed sensor, used in conjunction with a sensor on the rear RHS wheel, which was used for GIPSITRAC sensor synchronisation. o Microprocessor, which controlled the GIPSITRAC sensors and all I/O functions.
It was after the 2000 Sydney Olympics that the potential of the GIPSICAM system was truly appreciated within the RTA. Since then the GIPSICAM system has been used to routinely survey the state roads network within NSW. GCv2 attained operational status in 1999, and was decommissioned in mid-2006, after seven years of continuous service.
1.6 Why the need to upgrade GIPSICAM?
In December 2003, TMG International (Australia) “was engaged by RTA to undertake a review of the pavement condition data collection fleet, its condition and management, in order to assist RTA develop a strategy for replacement or enhancement of the data collection capability” (Roads and Traffic Authority, 2004 Strategy).
7 Introduction
Figure 4. The GCv2 digital video camera mounted on the roof of the vehicle
The results of the strategic review, released in July 2004 with the title “Strategy for Enhancement/Development of RTA Pavement Condition Monitoring Equipment”, found that GIPSICAM functionally provided a very important capability that needed to be maintained. However, operationally the review found that GCv2 was designed in an ad-hoc manner and nearing the end of its operational life, both of which were causing reliability issues. Relevant extracts from the report include (Roads and Traffic Authority, 2004 Strategy):
“The GIPSICAM vehicle is of concern. It would appear that it is not a sound choice for the duties required of it, having no facilities for operators and being close to the edge of its vehicular design performance. The vehicle has travelled a fairly high distance”. “GIPSICAM reliability is an issue because it is operated at the limit of its electrical capability”.
8 Introduction
“It clearly has electrical capacity issues, and the design has not considered the electrical loading. The alternator has been of concern, and thermal loading has led to lower reliability”. “The basic design has been undertaken on an ad-hoc basis. It is this design, with its inherent electrical loading that is the key reliability risk.” “There are no clear emergent developed alternative technologies available for procurement world-wide, and RTA technology appears as advanced as any currently in development elsewhere.” “The medium term solution will be the replacement of the vehicle with one capable of supporting the loading requirement reliably, i.e. with some capability to spare.”
Figure 5. The GCv2 DGPS and GPS antennae mounted on the roof of the vehicle
9 Introduction
The strategic review of the RTA pavement condition monitoring equipment was the catalyst for the building of GCv3. Armed with the report, Steve Greening, the GIPSICAM manager at the time, proceeded to enlist support and funding to undertake the development of the next generation GIPSICAM system.
Figure 6. The GCv2 computer/equipment rack in the cargo area of the vehicle
10 Introduction
1.7 Along came Dennis Entriken
Steve Greening was the “father” of the GIPSICAM technology within the RTA. In the early 1990’s Steve teamed up with Ken Root, two very clever surveyors who graduated from UNSW many years before, and together they pioneered MMS technology within the RTA. Others surveyors such as Robert Pierce, Chris Woodham and Roger Merritt also played a part in the early days of GIPSICAM. Steve developed and managed the GIPSICAM technology with the RTA. His official title was “Technology Development Specialist” within the Survey Services Section, Technical Services Branch of the RTA.
Steve Greening was nearing retirement and looking to pass the reins of GIPSICAM over to a new generation. Thus in 2003 Steve recruited an understudy to learn about GIPSICAM and to one day continue in his footsteps developing MMS technology within the RTA. That person was I, Dennis Entriken.
Figure 7. Mike Vernon, Steve Greening and Dennis Entriken
11 Introduction
My background skills and experience were broad, covering areas such as Computer Science, Software Engineering, GIS, Remote Sensing, Location Based Services, GPS Technology, Information Technology, Systems Integration and Project Management. For the previous 11 years I had worked mostly as a Software Engineer / GIS consultant for companies in Australia, the UK and Japan. Steve recruited me to be his GIPSICAM successor, realising that my unique combination of skills and experience were well suited to the task at hand - the development and management of the GIPSICAM technology within the RTA. In January 2005 approval was given by RTA senior managers to develop the 3rd generation GIPSICAM System within the RTA. My MPhil research topic was therefore to become “The development of the 3rd generation GIPSICAM system within the RTA”.
Research and development (R&D) efforts on GCv3 commenced in early 2005 and were completed in mid-2006. The first operational survey was on 4th April 2006 on roads in the Wollongong area. During the 2006 winter months more testing was conducted and procedures and validation mechanisms were developed in order to ensure the data collected was of best quality. GCv2 was decommissioned in mid-2006 and GCv3 officially commenced operations in September 2006 for the start of the 2006/2007 GIPSICAM survey season.
GCv3 has been utilised to successfully survey the NSW state road network for the last five GIPSICAM seasons, from September 2006 to March 2011.
During the period from 2005 to 2007 I took more than 4,800 digital photographs documenting the end-to-end GCv3 R&D process. I also amassed more than 9,000 files to support the GCv3 R&D process which included research papers, magazine publications, equipment and sensor specifications and user guides, and test data.
In November 2010, I was successful in winning the newly-created senior position of “Manager, Road Information Management & Integration Services” within the RTA. I am now responsible for managing the corporate spatial data and systems, developing real time traffic analytics platforms, and developing corporate EDW and BI capabilities.
12 Introduction
1.8 A brief overview of GIPSICAM v3 (GCv3)
Figure 8. The Roads and Traffic Authority’s GCv3 vehicle
The core components of the GCv3 vehicle are (See Figure 9): Mercedes-Benz Sprinter vehicle, with extensive after-market custom modifications, providing a reliable, spacious, comfortable, OH&S friendly and protective environment for both staff and equipment, so as to ensure safe, efficient and high quality surveys. DGPS receiver with real-time OmniSTAR VBS service corrections, which provides the primary absolute positioning of the vehicle trajectory. A backup GPS receiver, logging data that can be post-processed with GPS base station data to provide supplementary absolute position data, when required. It also provides navigation information to the vehicle operator.
13 Introduction
Fibre optic gyroscope (FOG), which provides the direction of travel of the vehicle, which in turn is used to determine relative positioning via dead reckoning. Laser-based optical sensor, which provides data on distance travelled. Digital readout odometers, which use the vehicle ABS pulses to determine distance travelled. Four IEEE 1394a mega-pixel progressive scan digital video cameras, with a selection of wide-angle, standard and telephoto mega-pixel lenses, which are used to collect high resolution digital video. Dual processor server, mounted in a custom-built vibration-dampened computer rack, which provides the computer processing power within the vehicle. Operator console consisting of a 15” flat panel monitor mounted on a custom- built monitor arm, an optical Marble Mouse and a flexible mini-keyboard. Two independent electrical/power systems, each with their own battery bank and alternator for charging, one electrical system for the vehicle and the other for the MMS equipment. GIPSITRAC inertial navigation system (INS), permitting relative position and vehicle attitude to be determined, comprising: o GPS receiver, used for GIPSITRAC sensor synchronisation. o Two micro-electromechanical system (MEMS) gyroscopes, used to determine the direction of travel and the horizontal radius of the vehicle path. o Two accelerometers, one placed in a longitudinal orientation and the other in a transverse orientation, used to determine the grade, cross-fall and vertical radius of the vehicle path. o Rotational speed sensor, used in conjunction with a manufactured 40- toothed steel cog fitted in-line with the driveshaft, used for GIPSITRAC sensor synchronisation and also provides data on the distance travelled. o Microprocessor, which controls GIPSITRAC sensors and all I/O functions.
14 Introduction
Figure 9. Overview of GCv3 sensor and equipment layout
Figure 10. GCv3 vehicle sensor block diagram
15 Introduction
The GCv3 vehicle surveys more than one third of the 17,623km of state road in NSW each survey season, from October through to March. Roads are surveyed in both directions. There are tight controls on image quality with procedures defining hours of data collection and direction of survey relative to the time of the day and the month of the year. The emphasis is on collection of high quality data, rather than high quantity (See Figure 11).
Figure 11. Front and side camera road images from GCv3
The data collected by the GCv3 vehicle is post-processed in the office to produce the following outputs: Road centreline vector data. Georeferenced “drive-along-the-road” imagery. Georeferenced road geometry data (grade, crossfall, horizontal and vertical radius).
Road images are captured every 10m along the road, in both directions (See Figure 12).
The processed data is then distributed/replicated immediately via the RTA’s wide-area network (WAN) to nine dedicated Novell severs at the larger RTA offices throughout NSW. Smaller offices utilise standalone Ethernet drives or external USB drives which have the data duplicated onto them at regular intervals. RTA staff members may then access the GIPSICAM data using RTA-developed software or commercial GIS software such as ESRI ArcGIS.
16 Introduction
Figure 12. GCv3 road image positions displayed on ©SKM orthophotography data
The GIPSICAM dataset consists of approximately 8,000,000 standard resolution front camera and side camera images occupying approximately 800GB of file space; approximately 8,000,000 high resolution front camera and side camera images occupying approximately 2.1TB of file space; and metadata, reference spatial data, and georeferenced orthophotography, topographic and digital street directory raster basemaps occupying approximately 1TB of file space. The complete dataset distributed to all GIPSICAM data servers covers all NSW state roads and occupies a total of approximately 4TB of file space. Historical GIPSICAM datasets are maintained on a single server and accessed by RTA staff members via the RTA intranet.
17 Introduction
1.9 Motivation
The use of a MMS to quickly collect large amounts of data has been proven to be a cost effective and very beneficial technology for Road Authorities around the world. The RTA recognised the benefits of utilising such a system in the management and maintenance of their state road network and have developed the GIPSICAM system to undertake that role.
A strategic review of RTA data collection undertaken by Sheldon Consulting for the RTA in 2003 highlighted the importance of GIPSICAM data and recommended an increase in the collection and use of GIPSICAM data (Roads and Traffic Authority, 2004 Strategy). A strategic review of RTA pavement condition monitoring equipment that occurred in 2004 highlighted reliability issues with GCv2. These reliability issues were attributed to the ad-hoc design of the system during its development over the previous 10 years (Roads and Traffic Authority, 2004 Strategy).
The strategic review of RTA pavement condition monitoring equipment also found that there was “no clear emergent developed alternative technologies available for procurement world-wide, and RTA technology appears as advanced as any currently in development elsewhere” (Roads and Traffic Authority, 2004 Strategy). The recommendation of the report was “the replacement of the vehicle with one capable of supporting the loading requirement reliably, i.e. with some capability to spare” (Roads and Traffic Authority, 2004 Strategy).
1.10 Research objectives
The aim of the project was to develop the third generation GIPSICAM system, utilising the experience from previous generation GIPSICAM systems, with the objective of designing a special purpose MMS that utilises current technology to provide a reliable, safe, efficient, accurate, high quality GIPSICAM capability to the RTA, for the purpose of rapid road asset data capture and other applications of MMS technology.
18 Introduction
1.11 Contribution of the research
The contribution of this research can be summarised as follows: The selection of a vehicle and its modification in preparation for use as a MMS were investigated. The selection of equipment and the integration of subsystems for use within a MMS were investigated. The calibration procedures, validation procedures, and operational procedures utilised by an operational MMS were investigated. The application of MMS technology within a Roads and Traffic Authority was investigated.
1.12 Outline of the thesis
This thesis consists of nine chapters.
Chapter 1 introduces the GIPSICAM technology and summarises the history of the previous generations of GIPSICAM, the new generation of GIPSICAM that was developed as part of this research, and describes why this research and development took place.
Chapter 2 briefly presents the results of trial outsourcing of the GIPSICAM capability in early 2006, describes some fundamental concepts regarding MMS equipment, and describes reference systems utilised within the RTA.
Chapter 3 describes the tasks undertaken to select and modify a vehicle for use as a MMS platform.
Chapter 4 describes the selection of equipment and the integration of subsystems for use within GCv3.
19 Introduction
Chapter 5 briefly describes the software utilised by GCv3.
Chapter 6 describes the calibration procedures, validation procedures, and operational procedures utilised by an operational MMS such as GCv3.
Chapter 7 presents some of the test data collected to verify that the GCv3 system was operating according to required specifications and producing high quality GIPSICAM data.
Chapter 8 presents an overview of the application of MMS technology within a Roads and Traffic Authority, including some case studies where the GCv3 system and/or GIPSICAM data was utilised as part of a solution.
Chapter 9 summarises the research and development activities associated with GCv3, recommends future research and development opportunities for GIPSICAM, and speculates on the future of MMS technologies within a Roads and Traffic Authority.
20 Review of MMS Technology
CHAPTER 2 REVIEW OF MMS TECHNOLOGY
2.1 Introduction
The earliest known MMS is the Aspen Movie Map, which was developed at MIT in 1978. It enabled a “virtual drive” through the streets of Aspen, Colorado in the USA. It featured four 16mm film, still cameras with wide angle lens that captured front, back and both side view images every 10 feet along the road. The capture of the images was triggered by an optical odometer attached to a bicycle wheel that was mounted at the rear of the vehicle. The images were matched to the road centrelines and stored on laserdisc for later viewing using a computer and a touch screen interface.
Research and development of MMS systems continued throughout the 1980’s and 1990’s at a number of educational and commercial institutions, with increased emphasis on quality and clarity of imagery and accuracy of georeferencing. Advances in video, navigation and sensor technologies, particularly in relation to high resolution digital video, the development of GPS, the integration of GPS and INS, and the development of cheaper more accurate MEMS and fibre optic sensors, have led to the development of commercial-grade MMS able to quickly, efficiently and accurately capture road assets along large road networks. Some notable MMS research and development included: MHIS (Mobile Highway Inventory System) developed by Alberta Transportation in the early 1980’s. See (Ross, 1983) and (Williams, 1986). GPSVan developed at Ohio State University in the early 1990’s. See (Bossler et al, 1991) and (Novak, 1993). VISAT (Video-Inertial-SATellite) developed at the University of Calgary in the early 1990’s. See (El-Sheimy, 1996). ARAN (Automatic Road Analyser) developed by Roadware in the early 1990’s. See (Maerz et al, 1999).
21 Review of MMS Technology
GIPSICAM (Global and Inertial Positioning Systems Image Capture for Asset Management) developed by the Roads and Traffic Authority in the 1990’s (See section 1.3 What is GIPSICAM?). Hawkeye developed by ARRB in the 1990’s. (See section 1.4 A brief history and overview of GIPSICAM v1 (GCv1)). (See Figure 13). See (ARRB, 2010) and (ARRB, 2011).
Since the development of the Aspen Movie Map, Roads and Traffic Authorities in Australia have recognised the potential of a Mobile Mapping System (MMS). The RTA developed the GIPSICAM MMS in the 1990’s and continues to operate a MMS vehicle. Queensland Main Roads also built a MMS in the 1990’s and continues to operate a MMS vehicle. Other Australian states outsource the road image data collection of their state road network to commercial MMS operators here in Australia, the most popular of which uses the ARRB Hawkeye 2000 system.
Figure 13. ARRB Hawkeye 2000 vehicle
22 Review of MMS Technology
2.2 Trial outsourcing of road image collection in FY05/06
In conjunction with the development of GCv3, the question was asked, “should this data collection be outsourced”? The response was a trial data collection exercise to ascertain the feasibility of outsourcing the data collection activity. When assessing the commercial land-based MMS technologies available in Australia in 2005, only one was considered potentially suitable for the NSW Roads and Traffic Authority (RTA) requirements: the ARRB Hawkeye 2000 system (See Figure 13). This was primarily because it has similar system components to GIPSICAM.
The road image data collection trial was undertaken in conjunction with the annual profilometry survey conducted by ARRB in early 2006. The results of this trial were deemed unsatisfactory due to the overall poor quality of the road images obtained. The issues though were not related to the in-vehicle technology, but rather the procedures undertaken in conducting the surveys and a conflicting requirement from the profilometry survey. Issues relating to the processing of the data was also a concern but are not discussed here. Extracts from a report by Dennis Entriken in December 2006 summarise the results of the ARRB road video data collection trial are presented below.
2.2.1 Background
ARRB were engaged during FY05/06 to collect road video image data as an extension to the profilometry survey. The following is a summary of the assessment of the visual quality of the recorded raw road video data.
23 Review of MMS Technology
2.2.2 Roads requested to be surveyed during FY05/06
A road video data collection survey plan was provided to ARRB prior to the commencement of survey in FY05/06.
Road # Name 1st Link # Last Lk # Region HW2 Hume Highway 0010 2346 Syd/South/SW HW6 Mid Western Highway 0010 0860 SW HW9 New England Hwy 0101 1550 North HW14 Sturt Highway 0012 0620 SW HW17 Newell Highway 2005 3431 SW/West HW20 Riverina Highway 1010 1461 SW HW22 Silver City Highway 0010 0275 West HW28 Carnarvon Highway 0110 0260 West MR57 Old Junee-Nyngan 0010 0160 SW MR61 Orange-Cobar 0010 0310 West MR61 Orange-Cobar 0615 0645 West MR70 Coolabah-Nr Hebel (Qld) 0015 170 West MR80 Narrandera-Mossgiel 0010 0311 SW MR86 Corowa 0100 0110 SW MR89 Tomingley-Narromine 0010 0080 West MR314 Yarrawonga-Mulwala 0010 0090 SW MR321 Jerilderie-Rankins Springs 0010 0170 SW MR410 Willanthry Br-The Priory 0011 0105 SW MR421 Cobar-Nr Bourke 0005 0140 West MR501 Lake Cargelligo-Booligal 0041 0071 SW MR501 Lake Cargelligo-Booligal 0505 0505 SW MR550 Corowa-Tocumwal 0060 0140 SW Table 1. Road video data collection survey plan (FY05/06) provided to ARRB
24 Review of MMS Technology
2.2.3 Roads surveyed by ARRB during FY05/06
Approximately 6430km of raw (unprocessed) road video survey data was delivered to the RTA, for all roads requested to be surveyed, as stipulated in the survey plan provided to ARRB (See Table 1).
2.2.4 Visual assessment of road video image quality
The road video image data collected by ARRB was checked and classified according to clarity, brightness, sharpness and general image appearance. The classification consisted of three ratings: 1. Acceptable (A) 2. Mostly Acceptable (MA) 3. Not Acceptable (NA)
Of the 6430km of road video image data collected by ARRB, 4030km were found to be of either of an Acceptable (A) or Mostly Acceptable (MA) visual quality, while 2400km were found to be of an unacceptable (NA) visual quality (See Table 2).
Assessment (visual) Length (km) Length (%) A + MA 4030 62.7 NA 2400 37.3 Total 6430 100 Table 2. Summary of road video image visual quality
2.2.5 Reasons for classifying road video image as unacceptable
Visual image quality is a subjective issue, however the general problems found with the ARRB road video images were related to motion blur, lighting, sun angle/direction and visual artefacts. Note that these problems are generally interrelated.
25 Review of MMS Technology
2.2.5.1 Motion Blur
Motion blur is the term used to describe blurry images caused by movement of the camera, or the subject, or both. In the case of mobile mapping technologies, motion blur is the result of a slow camera shutter speed with respect to the speed of the vehicle travelling along the road. The faster the vehicle is travelling, the faster the camera shutter speed required to capture a sharp picture. Solutions for motion blur include reducing the speed of the vehicle or increasing the camera shutter speed. If the camera shutter speed is increased, then the camera aperture will need to be opened wider to allow more light to enter the lens to ensure a correct exposure. Note however that opening the camera aperture wider will result in a reduced depth of field.
2.2.5.2 Lighting
Adverse lighting will result in images that are too dark or too light. Lighting problems are the result of incorrect exposure, caused by wrong camera settings or by light conditions outside of the range that can be handled by the camera (with respect to the current camera settings). Solutions for light related problems generally involve choosing more appropriate camera settings. Other solutions include using cameras (and lenses) with capabilities to handle a greater range of conditions, or by selecting a different exposure method (manual, shutter priority, aperture priority, automatic exposure, automatic gain). Finally, capturing road video image during daylight hours is mandatory.
2.2.5.3 Sun Angle/Direction
Collecting road video images with respect to inappropriate sun angles/directions will result in incorrect exposure or visual artefacts. These problems are the result of collecting data when driving towards the rising/setting sun resulting in glare and/or reflections on/off the lens itself, or any protective glass in front of the lens. Low sun angles also cause long shadows across the road pavement from trees and buildings
26 Review of MMS Technology
alongside the road, resulting in changing camera exposure levels and dark/light patches of road within the images. Solutions for sun angle/direction problems include driving in a direction away from the sun. Some generic rules are to avoid heading in a predominantly eastward direction in the morning, avoid heading in a predominantly westward direction in the afternoon, and avoid heading in a predominately northwards direction in early morning and late afternoon. Sunlight conditions are the most optimal for longer periods of time during the summer months of December/January when the sun is high in the sky.
2.2.5.4 Visual Artefacts
Visual artefacts are unwanted light/colours/glare/reflections that degrade the visual appearance of the images. Visual artefacts can be caused by camera hardware limitations such as “blooming”, where an intense light causes saturation of the digital camera pixels within the CCD, resulting in a vertical streak down the image. Another cause of visual artefacts is glare and reflections from the sun. Finally, dead insects on the protective camera glass also may produce visual artefacts. Solutions for visual artefacts include replacing the cameras with newer models which overcome problems such as “blooming”, not driving in a direction towards the sun, especially when the sun is low in the sky, and regular cleaning of the protective camera glass.
2.2.6 Coverage of roads requested in survey plan
Road video (GIPSICAM) users within the RTA expect complete coverage of a road during a single season. Of the 20 roads requested to be surveyed, only 6 roads were delivered where the majority of the road video data was at an “Acceptable” or “Mostly Acceptable” quality level. The total length of these 6 roads was 905km or 14% of the data delivered to RTA. These 6 roads were HW20, HW22, HW28, MR89, MR421 and MR550.
27 Review of MMS Technology
2.3 An overview of positioning and attitude technology utilised in GCv3
2.3.1 Global Positioning System (GPS)
The Global Positioning System (GPS) can provide the current position of the receiver anywhere on the earth, at any time, without delay. GPS was developed by the US Department of Defense with full operational capability declared in 1995 (USNO, 2011). GPS provides discrete, absolute positions with the level of accuracy/error in each position independent from each other.
2.3.1.1 Autonomous GPS
Autonomous GPS receivers are stand-alone instruments able to determine positions with an accuracy of the order of 8m (2D) or 16m (3D) using single point positioning. The effects of multipath, urban canyoning and poor satellite geometry can degrade the accuracy of autonomous GPS.
2.3.1.2 Differentially corrected GPS (DGPS)
The concept behind differential correction is straightforward. If the exact location of a static GPS is known then the aggregated errors associated with the satellite clocks, orbits and atmospheric delay can be determined and appropriate corrections calculated. These error corrections can then be “applied” to the GPS data from a GPS at an unknown location to mitigate the aforementioned errors and calculate a more accurate position. The GPS at the known location is generally referred to as the Base Station or Reference Station, and the GPS at the unknown location is often referred to as the Rover.
In general, the accuracy of differentially corrected GPS (DGPS) varies from 0.1m to 5m (2D) and 0.2m to 8m (3D) when using pseudorange measurements. However, the accuracy of DGPS depends on a number of factors such as the proximity of the Rover to
28 Review of MMS Technology
the Base Station and the accuracy of the known position of the Base Station. The effects of multipath, canyoning and poor satellite geometry on either the Base Station or the Rover receiver can also degrade the accuracy of DGPS.
2.3.1.3 OmniSTAR real-time DGPS
OmniSTAR provides a world-wide commercial real-time DGPS correction service based on Satellite Based Augmentation System (SBAS) technology. OmniSTAR has more than 100 reference stations around the world that are used to calculate the GPS error corrections, which are broadcast to the subscribers using communications satellites (OmniSTAR Pty Ltd, 2011). At the time of development of GCv3 OmniSTAR provided two commercial real-time DGPS correction services which were OmniSTAR VBS (Virtual Base Station) and OmniSTAR HP (High Performance). Yearly subscription to the OmniSTAR VBS service is approximately AUD$2500; very good value for money given the capability it provides. Once the subscription has been paid, then an “initialisation” procedure must be carried out in unison with the OmniSTAR service provider, after which real time corrections are received by the OmniSTAR-capable GPS and utilised to determine DGPS positions. The accuracy of the OmniSTAR VBS service is sub 1m (2D). The accuracy of the OmniSTAR HP service is 10cm (2D) and 20cm (3D). The datum currently used by the OmniSTAR DGPS service is ITRF2008.
2.3.2 Inertial Navigation System (INS)
Inertial Navigation System (INS) technology provides navigation information based on inertial forces detected by sensors. An INS can operate in GPS hostile environments such as in tunnels, urban canyons and along tree-lined state forest roads. When combined with an absolute positioning technology such as GPS, an INS can be used to determine relative positions and provide navigation correction information to ensure positional accuracy when operating in GPS hostile environments. However, an INS provides sequential, relative positions with the level of accuracy/error in each position dependant on the previous position resulting in the compounding of errors until such
29 Review of MMS Technology
time as a correction is made by utilising a more accurate position such as from GPS. The compounding error of subsequent INS positions is known as “drift”.
The performance of an INS can be categorised into three classes based on their rate of drift (measured in degrees per hour), which are rate grade (10 to 1000 deg/hr), tactical grade (0.1 to 10 deg/hr) and inertial grade (0.0001 to 0.1 deg/hr) (Kempe, 2011). Rate grade sensors are cheap but do not provide the levels of accuracy required by a MMS for capturing data in long tunnels, city CBD’s, state forests and national parks. Tactical grade sensors can now be procured easily and at reasonable prices, providing a cost effective solution to providing navigation information that meets MMS asset capture requirements (See section 8.3 Specific applications of the GIPSICAM data within the RTA).
2.3.2.1 Accelerometers
A linear accelerometer can detect acceleration (or deceleration) in a preset orientation or direction relative to the reference body on which the accelerometer is attached. If the orientation of the accelerometer is coincident with the forward direction of travel then acceleration in the forward direction can be measured. If the orientation of the accelerometer is perpendicular to the direction of travel, and thus coincident with a line from the left to the right of the vehicle, then acceleration in the left/right direction can be measured. Finally, if the orientation of the accelerometer is perpendicular to both forward motion and sidewards motion then acceleration in the vertical direction can be measured.
2.3.2.2 Gyroscopes
A gyroscope is a turn-rate sensor, which can be used to determine the direction of travel. There are three types of gyroscopes, which are “mechanical (a rotating mass suspended in gimbals), optical, and micro-electromechanical system (MEMS) vibrating structures” (Bonsen et al, 2005). MEMS gyroscopes employ the principle that if you
30 Review of MMS Technology
apply an angular velocity to a vibrating object then the resulting coriolis force can be used to determine the initial angular velocity (Avnet Kopp, 2003).
Figure 14. Basic principle of a vibrating gyroscope (Bonsen et al, 2005)
2.3.3 Odometer
An odometer is an instrument that measures distances travelled.
2.3.3.1 Variable Reluctance sensor
A “variable reluctance sensor placed above a rotating toothed steel disk generates a sinusoidal voltage signal from changing magnetic flux” (Bonsen et al, 2005). The sinusoidal voltage signal, or pulse, indicates that the steel cog has rotated to the next tooth, which indicates that a pre-calculated distance has been travelled.
31 Review of MMS Technology
Figure 15. Sensor signal generation in MR-sensors (Schmeißer et al, 1999)
Figure 16. Front/Side view of KMI sensor and steel cog (Schmeißer et al, 1999 and modified)
32 Review of MMS Technology
2.4 Reference Systems used in the RTA
Spatial data within the RTA is used in some form or another across all directorates of the RTA. Different directorates have different spatial data needs and utilise the GIPSICAM data in different ways. This has resulted in a number of distinct reference systems being used within the RTA.
The RTA utilises commercial software such as ESRI ArcGIS and in-house developed software such as GridLoc and the Location Translation Service (LTS) to reproject/transform data between the different reference systems used within the RTA.
Figure 17. RTA Lambert94 parameters
33 Review of MMS Technology
2.4.1 RTA Lambert94
RTA Lambert94 is a projection defined by the Roads and Traffic Authority (RTA) in NSW and adopted by the RTA in 1998. It is based on the Lambert Conformal Conic projection with the following parameters: The datum is GDA on the GRS80 ellipsoid. The two standard parallels, which define a secant projection, are 30.75o S and 35.75o S. The latitude origin is 33.25o S and the central meridian is 147o E. It has a false easting of 9,300,000 m and a false northing of 4,500,000 m. (See Figure 17). RTA Lambert94 is primarily used to display state-wide datasets within GIS software. The RTA Lambert94 projection was adopted by the NSW state government in 2006 and is now also known as GDA Lambert (LPMA, 2006).
2.4.2 Map Grid of Australia (MGA)
In the late 1990’s Australia adopted the Geocentric Datum of Australia (GDA) as our national datum and correspondingly adopted the Map Grid of Australia 1994 (MGA) as our national projection (GA, 2010; LPMA, 2010). MGA is based on a UTM projection, as was MGA’s predecessor the Australian Map Grid (AMG) that had realisations in 1966 and 1984. MGA is realised from the GDA datum, which utilises the Geodetic Reference System 1980 (GRS80) ellipsoid. There are eight UTM zones that cover Australia, which are zones 49 through to 56. Each zone is 6o wide with an origin at the intersection point of the central meridian and the equator. MGA has a false origin with a false easting of 500,000 m and a false northing of 10,000,000 m. The scale factor is 0.9996. MGA is primarily used by surveyors within the RTA. Three MGA zones cover NSW (LPMA, 2010): Zone 54 (central meridian 141o E) Zone 55 (central meridian 147o E) Zone 56 (central meridian 153o E)
34 Review of MMS Technology
2.4.3 RTA RoadLoc
The RTA RoadLoc reference system was developed in the 1980’s. The RTA needed a way to easily and accurately reference any part of any state road, but by way of a mechanism that did not require the purchase of expensive equipment or was unnecessarily complex and confusing. The solution was to develop a ‘1D reference system’ where a simple car odometer could be used to find a specified location.
The first simplification was to reference the road of interest. This is done via the roads unique road number consisting of seven digits, and padded with zeros at the start. However, because state roads are usually very long they are broken down into smaller sections of road called a ‘link’, which is represented by a four-digit number and padded with zeros at the start. The first link at the start of a road is link 0010. The link numbers are incremented by 10 in ascending order along the road. Lastly a ‘chainage’ (i.e. distance) from the start of a link can be used to specify a position on the road. This is the minimum information needed to specify a location on a road, however it does not provide information on direction of travel and does not take into account divided carriageways and ramps. Each link is given a carriageway code to specify the type of road being referenced. A carriageway version is also provided in case changes have been made to the link, which may cause reference problems. Lastly a direction of travel is provided in terms of Prescribed or Counter (the other way). Roads are classified in the Prescribed direction if they travel away from Sydney, or in a south to north direction or in an east to west direction.
35 Review of MMS Technology
The disadvantage of RoadLoc is that you need to know (or lookup) the road number of a particular road as well as the locations of each of the control points specifying the start of each link. The RoadLoc reference definition is: Road number* (7 digit integer) Link number* (4 digit integer) Link carriageway code (1 char) Link carriageway version (1 digit integer) Link chainage* (km with m precision) Direction of travel (P for Prescribed or C for Counter)
An example RoadLoc reference is: 0000001,0010,B2,0.100,P. The Red Cross in (Figure 18) indicates the actual RoadLoc reference location. This location can also be converted to RTA Lambert94 or MGA coordinates for different applications and different users.
Figure 18. RoadLoc reference: 0000001,0010,B2,0.100,P
36 Vehicle Selection and Modification
CHAPTER 3 VEHICLE SELECTION AND MODIFICATION
3.1 Introduction
The RTA has developed and operated MMS technology since the mid 1990’s. During that time the RTA has acquired a very good understanding of the operational requirements of a MMS vehicle. In particular it has learnt from past experience in terms of the mechanical, operational and functional aspects of maintaining and operating a MMS vehicle on Australian roads. The author was able to apply the prior knowledge and experience working with previous RTA MMS vehicles to the selection and modification of the new GCv3 vehicle. This chapter describes the issues addressed and actions taken in the selection and modification of the GCv3 vehicle.
3.2 Vehicle selection
Vehicle selection was an extremely important decision. The vehicle is the starting point for the development of the new MMS technology. It is like the clay used to create fine pottery or the foundations of a building. The selection of the vehicle can dictate the nature of the development of the MMS, hence a careful decision was needed.
The three primary considerations dictating the vehicle selection were quality/accuracy, efficiency and Occupational Health and Safety (OH&S).
The previous GIPSICAM vehicles had the cameras mounted externally to the vehicle that resulted in issues relating to the environment affecting the efficiency of the operation of the surveys. Cameras were exposed to dust, moisture, humidity, wind and temperature, which resulted in the need for regular equipment cleaning and maintenance during surveys, and occasionally resulted in equipment failure. Operational efficiency was affected. Possible solutions included externally mounted environmentally sealed cameras or externally mounted camera housings that could be environmentally sealed
37 Vehicle Selection and Modification
such as is used in scuba diving, however both possible solutions still suffered from exposure to temperature and were potentially expensive to procure, implement and maintain. What was needed was not a way to seal the cameras from the environment but a way to control the environment around the cameras, thus the decision was made to place the cameras inside the vehicle where the environment around the cameras could be controlled. Accordingly, the vehicle must be large enough to house the equipment.
Another aspect of the efficient operation of the surveys relates to the reliability of the vehicle in the field. Given that the RTA will only be maintaining one GIPSICAM vehicle, every day that the vehicle is out-of-service means one day lost from the survey season. The best way to ensure reliability is to procure a new vehicle and ensure that it is serviced according to manufacturer specifications. However, just because a vehicle is new does not mean that it is reliable, hence a historically reliable vehicle model should be selected. In addition, if the vehicle does break down in the field, for whatever reason, then it is important to minimise the out-of-service time and to be able to quickly restore the vehicle to operational status. This means that the vehicle needs to be field repairable by either the vehicle staff, for small problems, or by country mechanics for bigger jobs, and that parts should be easily obtainable. Thus the decision was made to procure a new vehicle that was common on Australian roads, which had a good reputation in terms of reliability, for which parts were easily obtainable locally within Australia, and which was easily repairable by local mechanics in the field if it was ever required.
To enable the collection of accurate, high quality data, the vehicle needed to be easy to drive and have minimal body roll. Thus vehicle options considered important were a short wheelbase, a rigid suspension, and an automatic transmission to minimise the effects of the vehicle suspension on the measuring of the pavement grade, crossfall and vertical radius, and to minimise the demands on the drivers’ ability to control the vehicle for the measurement of the pavement alignment and horizontal radius. The vehicle also needed an engine powerful enough to easily handle the weight of the vehicle and equipment, and a transmission capable of providing a smooth transition through the gears so as to minimise any additional effect of the operation of the vehicle on the measurement of the pavement grade, crossfall and vertical radius.
38 Vehicle Selection and Modification
Another important consideration in terms of quality was the height of the cameras. The decision to place all equipment inside the vehicle meant that the vehicle needed to be high enough to be able to mount the cameras inside the vehicle at a specified height from the ground. Previous RTA experience had found that a suitable height for the cameras was around 2.3m from the ground, which gives a good vantage point of the road. Thus the vehicle needed a high roof to support the internal mounting of a camera at a height above the ground of approximately 2.3m.
Occupational Health and Safety (OH&S) is am important consideration for the RTA. The safety of the vehicle staff and the other road users is paramount. Thus the vehicle safety rating was a consideration.
The decision to place all equipment in the vehicle meant that the vehicle must be large enough to house all of the equipment. In addition, spare parts, supporting materials and consumables should also be stored inside the vehicle and easily accessible. However, just being large enough to house the equipment, spares, supporting materials and consumables is not satisfactory in terms of OH&S. The vehicle must be spacious enough to house everything and still allow the staff to enter and exit the vehicle safely and comfortably. The staff should also be able to work with, maintain and service the equipment and access the spares, supporting materials and consumables without creating any OH&S issues such as having to lean inside the vehicle or crouch down for periods of time. The staff need to be able to work upright within the vehicle and should be able to walk around the equipment and storage containers without much difficulty. Thus the rear of the vehicle needed to high and spacious.
The equipment operator in the vehicle requires safe access to the equipment in the vehicle when the vehicle is parked off the side of the road or parked on the road shoulder. One way to accomplish this is for the equipment operator to access the equipment in the rear of the vehicle from inside the vehicle, such as moving from the front seats to the rear. However, this will be inhibited by the installation of a cargo barrier to protect the staff from the equipment in the rear of the vehicle in the event of a crash or an incident. The safest way to accomplish access to the equipment is for the equipment operator to enter the rear of the vehicle via a side door that is on the
39 Vehicle Selection and Modification
passenger side of the vehicle and away from any traffic. Thus the vehicle needed to have passenger side, side door access to the rear of the vehicle.
An equipment operator work area was to be constructed in the area around the passenger seat, involving the removal of the glove compartment and modification of the dashboard area around the glove compartment. Thus the vehicle should not have a passenger airbag. If the vehicle does come with an airbag then a request would be made to the manufacturer to not install the passenger airbag.
Other considerations included the potential of incorporating an additional vehicle alternator to charge the equipment batteries, and an additional air conditioner in the rear of the vehicle to assist in the control of the environment surrounding the equipment.
Based on the requirements for the new GCv3 vehicle a number of possible light commercial vehicles were proposed. The two vehicles short-listed were the Mercedes Sprinter Van (1st generation, Phase II) and the Iveco Daily Van (3rd generation). The RTA utilises both vehicles within its fleet, hence a wealth of first hand experience with the vehicles was obtained from the RTA staff that operated and maintained them. In addition, the GIPSICAM vehicle staff had extensive experience and expertise with both vehicles and were able to advise accordingly. After much debate and discussion, the vehicle model that was selected as the GCv3 vehicle was the Mercedes Sprinter 316 CDI (1st generation, Phase II).
In March/April 2005 the RTA Fleet Services procured a Mercedes Sprinter 316 CDI with a short wheelbase with the following options: high roof, sprintshift 6 speed automated manual transmission and air conditioning (See Figure 19). The manufacture of the new vehicle commenced in June 2005 at a DaimlerChrysler factory in South America. Shipment of the vehicle to Australia occurred in August 2005. The vehicle compliance plate is stamped August 2005. Delivery of the vehicle to the RTA Rockdale Works Office was on 13th September 2005.
40 Vehicle Selection and Modification
Figure 19. Mercedes Sprinter 316 CDI specification (captured from Mercedes-Benz, 2003)
In May 2005, just prior to manufacture of the vehicle, it was discovered that the request to not install the passenger airbag was not part of the specifications. To prevent delay in the delivery of the vehicle, the design of the equipment operator work area was altered so that the passenger airbag could be left operational.
41 Vehicle Selection and Modification
Figure 20. The new Mercedes Sprinter 316 CDI
After taking delivery of the vehicle in September 2005, and ensuring it was supplied as requested, modifications to customise the vehicle for its specialised purpose as a MMS commenced.
3.3 Vehicle modification
Planning the vehicle modifications started immediately after selection of the vehicle type and model. The period between March 2005 and September 2005 was spent looking closely at possible vehicle modifications, inspecting similar or example vehicle modifications, deciding what modifications were needed, selecting optimal modification solutions, planning the modifications, and finally procuring the services and parts to undertake the required vehicle modifications. The vehicle modifications to be undertaken were divided into four categories according to the skills and facilities required to perform the tasks.
42 Vehicle Selection and Modification
Figure 21. More pictures of the new Mercedes Sprinter 316 CDI
The first category was work that could be undertaken by RTA personnel.
The second category was for a general handyman/electrician/mechanic to undertake pre- modification and post-modification work, and to provide advice regarding technical issues. This service was procured on a time and materials basis and the successful service provider was K&M Consultancy Services Pty Ltd. K&M Consultancy Services provided Mr Mike Vernon, who had a range of technical skills including motor mechanic, electronics, carpentry, and fitting and turning. Mr Vernon also had extensive experience operating the previous GIPSICAM vehicle which proved extremely useful in terms of providing first hand experience operating MMS vehicles “out in the field”. He was also well versed on vehicle regulations, which proved to be very useful to the project.
The third category was for a specialist vehicle bodybuilder to undertake the majority of the modifications. A list of requirements was documented and services were procured
43 Vehicle Selection and Modification
via a RFQ (Request for Quote) process. The successful respondent was Emergency Transport Technology Pty Ltd, generally referred to as ETT. ETT had extensive experience customising Mercedes Sprinters, such as fitting them out as ambulances for the NSW Ambulance Service. The team at ETT were professional and ensured that the design and implementation of the different modifications were as per requirements.
The fourth category was for a specialist industrial design and engineering company to build and install a computer rack and monitor arm. A list of requirements was documented and services were procured via a RFQ process. The successful respondent was Industrial Evolution Pty Ltd. Industrial Evolution had extensive experience designing and manufacturing solutions for in-vehicle mounting of computers and monitors.
The modifications to the vehicle were undertaken during the period September to December 2005 and were conducted in four stages: The first stage consisted of preparing the vehicle for the bodybuilder and constructing the wooden equipment/storage cabinet in the rear of the vehicle. The second stage consisted of the bulk of the modifications undertaken by the vehicle bodybuilder. The third stage consisted of the construction and installation of the computer rack and computer monitor arm. The fourth stage consisted of the post-modification work in preparation for the installation of survey equipment and the implementation of an operational environment.
3.3.1 Equipment/storage cabinet
The vehicle needed to be self sufficient when out in the field. It is very inefficient in terms of productivity to have the vehicle staff driving around, particularly in remote localities where towns can be few and far between, looking for spare parts, replacement items and consumables. Thus a selection of commonly required spare parts, replaceable items and additional consumables are always stored in the vehicle in case they are ever
44 Vehicle Selection and Modification
needed. In addition, tools and equipment need to be stored and housed in secure locations.
Ideas to solve the issue of storage included the use of storage bins, cardboard boxes and plastic containers to store equipment, parts, items and consumables. However, this would result in untidy, disorganised storage that may end up moving around in the back of the vehicle. Securing the bins and containers was also a possibility but also a “messy” solution. A wooden equipment/storage cabinet was suggested and adopted as the solution to this problem. Different compartments could be built for tools, spares, equipment, consumables, etc., providing an organised storage area and a secure housing area for equipment.
The cabinet was designed to have storage areas at the rear of the vehicle that can be easily accessed via the rear doors, storage areas on the top of the cabinet which can be accessed from both the rear of the vehicle and inside the vehicle, and the secure equipment housing area at the front of the cabinet so the equipment may be accessed from inside the vehicle (See Figure 22).
Figure 22. Rear view of cabinet showing disassembled storage area
The rear of the cabinet contains six storage areas; three that are sliding drawers with rails and a locking mechanism (See Figure 23). Cabinet rear compartment No. 1 is the storage area for a 10 litre metal “Jerry Can”, to carry extra diesel for remote survey work. This is accessed via the top of the cabinet (See Figure 24). Cabinet rear
45 Vehicle Selection and Modification
compartment No. 2 is the storage area for tools used to conduct maintenance or repairs on either the vehicle or the equipment. Cabinet rear compartment No. 3 is the storage area for consumables such as blank DVDs and chalk, and reference materials such as road maps and accommodation guides. Cabinet rear compartment No 4 is the storage area for spare vehicle parts and spare electrical and equipment items. Cabinet rear compartment No. 5 is the storage area for the second spare tyre. The first spare tyre is stored under the rear of the vehicle. The additional spare tyre helps mitigate the risks associated with staff safety and loss of productivity in the event of getting two flat tyres in a remote area that could be a long distance from the nearest town. Cabinet rear compartment No. 6 is the storage area for the informative, warning and regulatory magnetic signs that are mounted on the vehicle during surveys to warn other road users of the MMS survey operation when in progress.
Figure 23. Rear view of cabinet showing five storage compartments
Figure 24. Rear and top view of cabinet showing diesel “Jerry Can”
46 Vehicle Selection and Modification
Figure 25. Rear view of cabinet showing drawers
The top of the cabinet contains eight storage areas; six that are secure, one that is out of sight, and one that is a flat area for general storage. Cabinet top compartment No. 1 is the storage area for the vehicle Sharps Removal Kit. This kit contains safety equipment for dealing with sharp objects such as syringes that may be found in public areas. Cabinet top compartment No. 2 is the storage area for the vehicle First Aid Kit. Cabinet top compartment No. 3 is the storage area for the ruggedised briefcases that securely house the external hard drives that are used to store the terabytes of video and sensor data that is collected during surveys. Cabinet top compartment No. 4 is the storage area for the documentation carried in the vehicle. This includes technical/operational/quality procedures, checklists, survey run sheets, Material Safety Data Sheets, Safe Working Method Statement, vehicle and parts manuals, equipment manuals, and other related documentation. Cabinet top compartment No. 5 is the storage area for fresh drinking water. The vehicle is often working in remote areas and it is important that fresh drinking water is carried as a safety precaution. Cabinet top compartment No. 6 is the storage area for the vehicle rubbish bin. Cabinet top compartment No. 7 is the storage area for the 20-metre electrical extension lead that is utilised to connect the vehicle to 240 volts mains power when required. This storage area is located underneath cabinet top compartments No. 1 and 2, and is accessed via either of these compartments. Cabinet top compartment No. 8 is not a secure area but is rather a flat area for general storage. It is covered by carpet to provide a non-slip surface. General items such as suitcases or boxes can be placed there.
47 Vehicle Selection and Modification
The front of the cabinet contains seven storage areas; five that are secure, and the other two that are utilised to mount panels. Cabinet front compartment No. 1 is sealed and has a 240 volts safety switch mounted on the front. The safety switch provides protection from electrocution while the vehicle is connected to the 240 volts mains power. Cabinet front compartment No. 2 is the secure mounting area for the Fibre Optic Gyro and the battery charger. Cabinet front compartment No. 3 is the secure mounting area for one of the two batteries used to power the 12 volt equipment circuit in the rear of the vehicle. Cabinet front compartment No. 4 is the secure mounting area for the GIPSITRAC box. The location of this compartment is the centre of the vehicle and over the rear axle. Cabinet front compartment No. 5 is the secure mounting area for the second of two batteries used to power the 12 volt equipment circuit in the rear of the vehicle. Cabinet front compartment No. 6 is the secure mounting area for the 12 volt power conditioner. Cabinet front compartment No. 7 is currently not utilised other than to mount the on/off switch for the 12 volt power conditioner.
Figure 26. Front and top view of structural walls of the cabinet
48 Vehicle Selection and Modification
Figure 27. Front view of the cabinet showing the seven storage areas (from L to R)
3.3.2 Cargo barrier
Whenever equipment, tools or items are placed in the rear of a vehicle there is always the risk if involved in an accident or if braking suddenly that objects may fly forward and injure the vehicle staff at the front of the vehicle. This is an OH&S issue.
The solution is to place some protection or a barrier between the staff in the front of the vehicle and the equipment in the rear of the vehicle. Cargo barriers are designed specifically to accomplish this task and do so very well. The issue however was that the video cameras were to be mounted on a shelf, up high, at the front of the vehicle (See section 3.3.6 Camera turret). If a cargo barrier to suit a high roof Sprinter was installed then the vehicle staff would not be able to access the cameras from the cargo area of the vehicle as the cargo barrier would reach to the roof. If a cargo barrier to suit a low roof
49 Vehicle Selection and Modification
Sprinter was installed then there may have been a gap between the top of the barrier and the camera shelf, which would be a safety risk. The solution was to procure a cargo barrier to suit a high roof Sprinter and to “cut it down” to fit. ETT was able to modify the cargo barrier, certify its use as a cargo barrier and install it securely in the vehicle.
A clear plastic sheet was fixed to the cargo barrier to help separate the environment in the front of the vehicle from the environment in the cargo area of the vehicle (See section 3.3.3 Air conditioning). However, an issue was found where the clear plastic sheeting on the cargo barrier was catching the wind and vibrating when the driver’s/passenger’s window was down and when travelling at highway speed. This issue was solved with a combination of applying Silastic between the cargo barrier mesh and the clear plastic sheeting, and implementing a procedure to drive the vehicle with the windows closed when travelling at highway speeds.
Figure 28. The cargo barrier being manufactured (L) and installed (R)
3.3.3 Air conditioning
One of the issues encountered when operating a MMS in the remote western areas of NSW during the summer months is heat. The temperature outside the vehicle can reach 45 degrees Celsius. In addition, computers and equipment generate heat while operating so the inside of the vehicle can become very hot. Equipment such as computers, batteries and hard drives do not operate efficiently, and may even malfunction when exposed to such high temperatures. Likewise, staff do not like to work in such extreme
50 Vehicle Selection and Modification
conditions. However the big risk for the vehicle staff is dehydration if they do not keep cool and keep hydrated, which is an OH&S issue.
The simplest solution is to schedule surveys in the far west of the state during cooler months. The problem is that the best time to conduct surveys is in the summer months when the sun is high in the sky and directly above the vehicle. Even at the start and the end of the survey season, October and March respectively, the temperature in remote western areas of NSW can reach 39 and 41 degrees Celsius respectively. The solution was to install air conditioning in the vehicle.
The vehicle arrived with the factory-installed air conditioner. However the vents for the air conditioner were in the front of the vehicle, which means the cargo area of the vehicle would not have adequate ventilation from the front air conditioner and this area of the vehicle would still get hot. The solution was to install an additional air conditioner in the cargo area of the vehicle. It would be installed under the roof at the very rear of the cargo area, with the air vents facing forward. However, the problem with this solution was that it was not possible to install both a second air conditioner compressor and a second alternator (See section 3.3.5 Equipment electrical system). The advice obtained from ETT was that both air conditioners could be operated from the one OEM compressor, which is what was implemented. ETT installed a second air conditioner in the position and orientation specified, and ran it off the OEM compressor. A second set of air conditioner controls was installed on the centre dash to control the fan level for the air conditioner in the cargo area.
Figure 29. The 2nd air conditioner being installed (L) and when finished (R)
51 Vehicle Selection and Modification
Having installed the second air conditioner it was found that when the vehicle staff entered and exit the vehicle the cold air from the cargo area would escape. In order to minimise the fluctuations of temperatures in the cargo area, a clear plastic sheet was fixed to the cargo barrier (See section 3.3.2 Cargo barrier). This enabled different temperatures to be set in the front of the vehicle and in the cargo area of the vehicle.
3.3.4 Cargo area insulating and soundproofing
Operating a MMS in remote areas of NSW during the summer months exposes the vehicle, staff and equipment to high temperatures. As described above, the vehicle was procured with a factory fitted air conditioner in the front of the vehicle, and a second air conditioner was installed in the cargo area of the vehicle to circulate cool air throughout the vehicle. In addition, a clear plastic sheet was fixed to the cargo barrier to separate the environment in the front of the vehicle from the environment in the cargo area of the vehicle (See section 3.3.2 Cargo barrier) to help keep the cool air inside the vehicle. However, heat from the outside sun and wind can still enter the vehicle via the walls and roof of the vehicle. The solution was to insulate the walls and roof of the vehicle.
Computers and equipment generate noise, such as the hum from a cooling fan. In addition, noises from the outside road environment can propagate inside the vehicle. In the cargo area, where the walls are hollow and the area is essentially a giant “drum”, noises are amplified and can reverberate inside the vehicle. The solution was to soundproof the walls of the vehicle.
Firstly, the cavities in the walls of the cargo area of the vehicle were filled with foam rubber. This was to soundproof the walls and add a layer of insulation. Next, the walls and roof were covered in a thin foam rubber to provide insulation. Moulded sheet metal panels were then placed over the insulating foam rubber and fixed in place. Finally, a vinyl covering on the walls and roof provided an aesthetic finish to the modification.
52 Vehicle Selection and Modification
Figure 30. Installation of the cargo area insulating and soundproofing
3.3.5 Equipment electrical system
Many of the issues that occurred out in the field with the previous GIPSICAM vehicle were related to the vehicle’s electrical system. The alternator struggled to provide charge to the battery, particularly in hot conditions when the efficiency of the alternator was lower. The vehicle and equipment consumed more power than the battery could provide, meaning equipment would not operate normally or would stop working. It was often the case that productivity was compromised due to power issues.
The new vehicle needed to have the capacity to power both the vehicle systems and the equipment, and still have sufficient spare power capacity for the installation and operation of additional equipment in the future.
Initial plans were to install an extra-heavy-duty battery in the vehicle, but it was quickly found that the current capabilities of wet cell car batteries would not be able to provide the capacity required. Multiple batteries could be wired in parallel to maintain the voltage but increase the capacity. However, where would the additional batteries be mounted? The engine bay of the Sprinter did not have any room to accommodate additional batteries. There was room underneath the vehicle where additional batteries could be mounted but there they would be exposed to the environment, and potentially susceptible to damage from rocks flying up under the vehicle. The other possibility was to place the batteries inside the vehicle where they could be protected and the
53 Vehicle Selection and Modification
environment could be controlled. Placing the batteries inside the vehicle presented a challenge as wet cell batteries need suitable ventilation in case hydrogen gas is produced if the batteries are overcharged. Wet cell batteries also need to be kept upright and would be inconvenient to maintain, having to remove the batteries to check and service them. Thus the option of using sealed (valve-regulated) batteries was explored and found to be the best solution. Sealed batteries require no maintenance, and present less risk of exposure to the chemicals and compounds inside the batteries. Heavy duty 12V and 24V sealed batteries are commercially available. The next question was, what type of battery was best suited to use as an equipment power source? Standard “starting” type batteries were not suitable as they are designed to provide quick bursts of energy. The type of battery required was a “deep cycle” battery, designed to provide a continuous flow of energy. Further investigation of different sealed lead acid battery types revealed three types: wet cell, Gel cell, and AGM (Absorbed Glass Mat). Wet cell technology has a shorter lifespan than Gel cell and AGM batteries, and do not hold their charge well when not in use. Gel cell and AGM are both well suited to deep cycles applications, however the Gel cell batteries are more prone to premature failure from overcharging. Therefore, the battery requirements of the equipment electrical system were multiple, heavy duty, sealed, maintenance free, deep cycle, AGM batteries. The voltage was yet to be determined.
Studying other vehicles that have had additional non-standard equipment installed revealed several interesting observations. One idea, which was observed in a bus that had been converted to a mobile home, was to have a separate electrical circuit for the equipment and appliances. Having two independent electrical circuits meant that the equipment circuit could be isolated from the vehicle system circuit and allowing flexibility in the design of the equipment circuit with no dependencies from the vehicle system circuit. For example, the GIPSICAM vehicle system electrical circuit was 12V but the equipment circuit could be designed to operate at 24V, and hence truck batteries could be used to provide greater power capacity.
Given the power issues with the previous GIPSICAM vehicle, and keeping in mind the goal of operational efficiency in the field, the idea of having a redundant means of powering the vehicle systems and the equipment was explored. If a 12V circuit was
54 Vehicle Selection and Modification
implemented for the equipment circuit, instead of a 24V circuit, then the 12V vehicle circuit and the 12V equipment circuit could be a redundant backup circuit for the other.
Having an independent circuit for the equipment raises another question. How will the batteries of the equipment circuit be charged? The bus that was converted to a mobile home had a 12V/240V equipment/appliance circuit and a 24V vehicle circuit. The 12V/240V equipment/appliance circuit was connected to a 12V wet cell car battery secured in a compartment that was accessible from the outside of the bus. The 12V battery was charged via a battery charger that was connected to external 240V mains power via a power switch and a circuit breaker. Hence each night when the bus was parked at a powered campsite or at a caravan park, the 12V battery could be charged via 240V mains power. In addition, a petrol-powered generator was also installed in the bus so if the bus was parked at an unpowered campsite then the 12V battery could be charged via the battery charger connected to the generator. This is a satisfactory way to keep the equipment/appliance circuit 12V battery charged given the usage pattern of the bus, however this is not the modus operandi of the GIPSICAM vehicle. The equipment in the GIPSICAM vehicle needed to be operational all day long so surveys could be conducted. One option was to install a generator in the GIPSICAM vehicle to charge the equipment circuit batteries, but finding a place to securely install the generator and then considering the issues of operating the generator all day, every day, during surveys suggested that this was not an optimal solution. Another option was to install a bank of batteries wired in parallel with sufficient capacity to operate the equipment all day long, but this is a rather extreme approach and would require that all of the batteries be recharged over night. Thus a way to charge the equipment circuit batteries, without the use of a generator, while operating the vehicle, was needed. Another option was to connect the equipment circuit batteries to the vehicle alternator, but the risk was that the existing alternator may not be able to provide the charge required for both electrical circuits on a continuous basis. A heavy-duty alternator was another consideration, but it still suffered from the lack of redundancy inherent in using a single means of charging all batteries in the vehicle. The best solution was to install an additional alternator, a heavy-duty model, to charge the equipment circuit batteries. A second alternator would solve the issue of keeping the equipment circuit batteries fully charged while the vehicle was conducting surveys. A second alternator also provided a level of redundancy. If a
55 Vehicle Selection and Modification
distribution switch is wired into the vehicle then the two alternators could be utilised to charge independent electrical circuit batteries, or if one alternator failed then the other alternator could be used to charge both electrical circuit batteries so that surveys could continue until the next town was reached and the faulty alternator was either repaired or replaced. An external power regulator could be used to regulate the charge from the second alternator and thus ensure optional performance in terms of charging the equipment circuit batteries.
The next consideration was in regards to the supply of voltage to the equipment. Both 12V power and 240V power needed to be available. Equipment and appliances operating at 12V could be connected to the 12V battery via a power conditioner unit. Equipment and appliances operating at 240V could be connected to the 12V battery via a power inverter unit. Both 240V mains power and 240V generator power could be connected to 240V equipment and appliances (in the example of the bus) via a circuit breaker. The decision was made that a generator was not going to be installed in the GIPSICAM vehicle, but otherwise all of the other solutions such as the 12V power condition, the power inverter and the 240V mains circuit breaker were suitable for inclusion in the equipment electrical system. All 240V power would be available via standard 240V wall sockets that would be mounted in the vehicle. All 12V power would be available via a bank of custom-built 12V power sockets.
There would be two separate, independent electrical circuits for the vehicle system and for the equipment. The voltage of the equipment circuit would be 12V to be compatible with the vehicle system circuit so that the two could be used as a backup for the other and hence provide redundancy. Two heavy duty, sealed, maintenance free, deep cycle, AGM, 12V batteries were to be procured and installed. The battery type selected was the Lifeline GPL-31T (AGM, 12V, 105Ah, 600CCA). Two Lifeline GPL-31T batteries were procured and securely installed within the designated compartments of the equipment/storage cabinet. The batteries were wired in parallel into the equipment electrical system, providing a total of 210Ah at 12V.
56 Vehicle Selection and Modification
Figure 31. The two Lifeline GPL-31T batteries were wired in parallel to provide 210Ah
The batteries were to be charged by a dedicated heavy-duty alternator. The alternator needed to be of a brand and model that was commonly used within rural NSW so that it could be repaired or replaced in small country towns by local mechanics. The alternator selected was the Bosch K1 9-120-060-042 (12V, 120A), designed for John Deer agricultural, industrial and marine applications. The Bosch K1 alternator comes fitted with the Bosch Internal Regulator W080-29N, however the internal regulator was removed and replaced with a Bosch Brush Holder 101020 so that a 3rd party regulator could be utilised. A Bosch High Airflow Alternator Fan 101003 was fitted to the alternator to keep the alternator cool. A Bosch Two Groove Alternator Pulley was fitted to the alternator to eliminate the risk of belt slippage. Bosch alternators need to be operated at speeds of 6000 RPM or greater to achieve their rated output, thus the selection of pulley ratio verses vehicle operating speed is critical to ensure optimal operation of the alternator. However, the operating speed of the GIPSICAM vehicle was to be either 80 km/h or the posted speed limit, whichever is the lower of the two. In urban areas the vehicle will be utilising lower gears and often operating in stop/start
57 Vehicle Selection and Modification
traffic, hence an optimal ratio was difficult to determine. After consultation with a number of experts the decision was made to use a pulley ratio of 2.5 to 1. This effectively meant that the alternator would be operating at the rated output of 120A when the vehicle engine was operating at or above 2400 RPM. Thus measuring the diameter of the crankshaft pulley and calculating a ratio of 2.5 to 1 determined the diameter of the Bosch Alternator Pulley. The alternator was installed on the lower driver-side side of the engine block, attached to the existing OEM alternator via a bolt and a tensioning bracket.
Figure 32. The engine bay being disassembled (L) and the Bosch alternator (R)
It was decided to not use the Bosch alternator internal regulator. A 3rd party external “smart” regulator would be used instead. The reasons for using an external “smart” regulator are that it could deliver microprocessor controlled, temperature compensating, multi-step charging and allows the fine-tuning of the charging parameters to optimise the charging and longevity of the batteries. The “smart” regulator selected was the Ample Power Next Step Regulator NEXT-12P (12V, P-Type Alt). The NEXT-12P was installed in the cavity underneath the passenger-side front seat.
58 Vehicle Selection and Modification
Figure 33. The Ample Power Next Step Regulator NEXT-12P
The decision to use two separate electrical circuits for the equipment and the vehicle means that a form of redundancy now exists in terms of the electrical systems. If the vehicle battery or the vehicle alternator should fail then the equipment batteries and the equipment alternator can take over the generation and supply of power for both the equipment and the vehicle systems. Alternatively, if the equipment batteries or the equipment alternator/regulator should fail then the vehicle battery and the vehicle alternator can take over the generation and supply of power for both the equipment and vehicle systems. However, a mechanism to enable the “switching” between the electrical systems was required, use of a power distribution switch was decided. The power distribution switch would be wired to allowed the vehicle staff to operate the electrical systems in one of three ways: each alternator charging its respective batteries and providing power to the respective systems, the vehicle alternator charging all batteries and providing power to both systems, and the equipment alternator charging all batteries and providing power to both systems. The power distribution switch selected and installed was the Hella Dual Battery Master Switch 2767 (6V – 36V, 310A/500A at
59 Vehicle Selection and Modification
12V). The distribution switch was mounted on the rear of the passenger seat and is accessed by the vehicle staff by reaching behind the passenger seat and rotating the switch to select the desired setting.
Figure 34. The distribution switch (L) and its installation on the passengers seat (R)
Having two separate electrical systems that can be utilised as a redundant power supply for the other via a distribution switch means that if ever there was a problem with one of the power supplies then the other power supply can take over and allow the vehicle to continue operation until the next town was reached where repairs could be undertaken. However, if the equipment batteries go flat while surveying the road then the computer will fail and the current survey data will be lost. The solution adopted was to install a Dual Battery Monitor (Digital Voltmeter) that allows the vehicle driver to monitor the voltage of both circuits and take precautionary actions should it be noted that one of the circuits started to drop in voltage. The Dual Battery Monitor was installed in the lower right-hand-side of the vehicle dashboard, relative to the driver’s seat.
The equipment circuit was to have a number of 12V devices and sensors connected to it and drawing power. However, as a 12V battery expends power, the voltage supplied by the battery will drop. In addition, the voltage supplied will increase when charged by the alternator. However, electrical equipment is designed to operate optimally at a specific voltage. Thus a way was needed to ensure a constant voltage was supplied to the equipment. Additionally, precautions were needed to ensure that the equipment was not subjected to interference, backfeed emf, voltage spikes, surges or transients from the power supply. The solution was to install a 12V power conditioner. The interVOLT
60 Vehicle Selection and Modification
SPCi21220 (10V – 16V, 12.5V/13.6V, 20A/25A) was selected. The 12V power conditioner was installed in the rear cargo area equipment cabinet (See section 3.3.1 Equipment/storage cabinet) and supplies 12V power from the two deep cycle AGM 12V batteries to the 12V equipment in the vehicle.
Figure 35. The interVOLT power conditioner
The flexibility to utilise standard mains power equipment such as a soldering iron or a desktop computer in the vehicle was seen as an important capability. Thus the equipment circuit also needed the capability to supply 240V AC. Many campervans and mobile homes, such as the bus that was converted to a mobile home, have a power inverter installed that enables the use of standard household equipment such as a TV to run off a 12V battery. A power inverter can convert 12V DC to 240V AC. The power inverter selected was the Sinergex PureSine 1500 Series II PS2-1500-212 (12V, 1500W). The power inverter was installed on the bottom shelf of the computer rack (See section 3.3.7 Computer/equipment rack). A surge and overload protected power
61 Vehicle Selection and Modification
board was plugged into the power inverter that provides the 240V AC power supply for the vehicle.
Figure 36. The Sinergex power inverter
It was envisaged that the equipment in the vehicle might sometimes be operated while in the workshop, particularly when performing testing, maintenance and some calibration procedures. However, operating equipment in the vehicle can quickly run- down a 12V battery if the vehicle engine is not running and providing charge from the alternator to the batteries. This is more so the case with 240V equipment powered via a power inverter. The equipment had to be operated while the vehicle was not running its engine, but without running down the batteries. The solution was to incorporate a combination of external 240V AC mains power and charge the batteries using the external power. This solution also provided a limited amount of redundancy as it could be used as a way to recharge the batteries in the vehicle overnight if there were alternator problems during the day, and hence leaving the batteries at less than 100% charge. The external mains power would be connected via an extension cord, from a 240V power socket outlet on the workshop wall, to a female power socket on an outside panel at the rear of the vehicle on the driver’s side above the wheel arch. The female power socket selected and installed was a Clipsal Power Inlet IP34. The female power socket on the outside of the vehicle was connected to a safety switch mechanism inside the vehicle, the device selected was a Clipsal MDRC Housing (an electrical equipment rack with a DIN rail for mounting Modular DIN Rail Components) with the following MDRCs installed: ABB E 271 63A 240V AC Main Switch, ABB DS951-AC C10 Safety Switch (10A, 6kA, RCBO), ABB S 231 C 10 Circuit Breaker (10A, 3kA, MCB),
62 Vehicle Selection and Modification
ABB SH 201 L C10 Circuit Breaker (10A, 4.5kA, MCB). The Main Switch can turn on/off 240V AC mains power to the vehicle. The Safety Switch protects against current leakage to earth, overloads and short-circuits.
Figure 37. External power connector (L) and Clipsal MDRC Housing (R)
The 3kA Circuit Breaker MDRC installed in the Clipsal MDRC Housing protects a 240V AC circuit running to a double power socket that is mounted on the outside, bottom, rear of the computer/equipment rack. When the vehicle is connected to 240V AC mains power, the power board supplying power to the 240V AC equipment in the vehicle can be unplugged from the inverter and plugged into the double power socket to enable the operation of the 240V AC equipment without the need to use the power inverter.
The 4.5kA Circuit Breaker MDRC installed in the Clipsal MDRC Housing protects a 240V AC circuit running to a 12V deep cycle battery charger. The deep cycle battery charger enables the “smart” charging of the equipment batteries when the vehicle is parked in the workshop or at a motel.
To distribute the 12V power supply, a small power supply distribution box was developed. The design consisted of an inline replaceable fuse, a 4-pin male connector socket with a screw-on securing mechanism, and an on/off switch. A corresponding 4- pin female connector was soldered onto the power cable of any 12V device that was to be used in the vehicle. A device could then be quickly and easily connected to the 12V power supply. The screw-on securing mechanism ensures that the connection does not
63 Vehicle Selection and Modification
come loose. The 12V power supply distribution box was mounted in the bottom shelf of the computer/equipment rack.
For devices that require a 5V power supply, two 12V to 5V transformers were built and installed in the computer/equipment rack.
Figure 38. 12V power supply (L) and 5V power supply (R)
To provide some additional protection against power failure and loss of data, a UPS was installed in the vehicle. The advantage of a UPS is that if the batteries in the rear of the vehicle do go flat then the UPS can provide a couple of minutes power to shutdown the system and save the current survey data. In addition, a UPS could be used to enable the switching of internal 240V power from the inverter to 240V mains power or back the other way, so the equipment does not need to be shutdown and restarted when leaving the depot in the morning or returning to the depot in the afternoon if the equipment operator is doing other work for an extended period of time before/after.
In summary, the components of the equipment electrical system, as constructed, consisted of: Independent 12V circuit. Two Lifeline GPL-31T batteries (AGM, 12V, 105Ah, 600CCA) wired in parallel to provide 210Ah at 12V. A Bosch K1 9-120-060-042 alternator (12V, 120A) with the Bosch Internal Regulator W080-29N removed and replaced with a Bosch Brush Holder 101020.
64 Vehicle Selection and Modification
A Bosch High Airflow Alternator Fan 101003 and a 2.5 to 1 ratio Bosch Two Groove Alternator Pulley fitted to the alternator. An Ample Power Next Step Regulator NEXT-12P (12V, P-Type Alt.). Hella Dual Battery Master Switch 2767 (6V – 36V, 310A/500A at 12V) wired to allow independent and redundant operation of the equipment circuit and the vehicle circuit. A Dual Battery Monitor (Digital Voltmeter) to display the voltage of both electrical circuits. An interVOLT SPCi21220 power conditioner (10V – 16V, 12.5V/13.6V, 20A/25A). A Sinergex PureSine 1500 Series II PS2-1500-212 (12V, 1500W) power inverter to convert 12V DC to 240V AC. A Clipsal Power Inlet IP34, connected to a Clipsal MDRC Housing with the following MDRCs installed: ABB E 271 63A 240V AC Main Switch, ABB DS951-AC C10 Safety Switch (10A, 6kA, RCBO), ABB S 231 C 10 Circuit Breaker (10A, 3kA, MCB), ABB SH 201 L C10 Circuit Breaker (10A, 4.5kA, MCB). A 12V Deep Cycle Battery Charger. Two 12V to 5V transformers. A 12V power supply distribution box. A 240V AC surge and overload protected power board. UPS.
3.3.6 Camera turret
Previous generations of the GIPSICAM vehicle had the cameras mounted externally to the vehicle, leaving the equipment exposed to the environment. This exposure to heat/cold, moisture and dust caused equipment failure and operational issues on a regular basis, which resulted in reduced productivity. For the 3rd generation GIPSICAM vehicle the decision was made to install the equipment inside the vehicle where the environment could be controlled. The next question to be answered was, how many cameras should be installed and in what configuration?
65 Vehicle Selection and Modification
The 2nd generation vehicle supported a forward-facing camera with a wide angle lens, mounted 2.3m above the ground, mounted in the centre of the vehicle, facing the direction of travel, and pointing down at the road at an angle of 4° below the horizontal. For compatibility reasons it was decided that the new vehicle would also incorporate a forward-facing camera with a wide angle lens utilising the same mounting characteristics as the previous vehicle (See Figure 39 and Figure 40). This camera is known as the “front camera”. The front camera would be the default camera for all standard GIPSICAM surveys.
Roadside assets are often located off the road carriageway, or adjacent to the road. Consulting internal clients and users revealed that video of these assets would potentially be useful. Thus it was decided that a 2nd camera would be incorporated into the design of the new GIPSICAM vehicle. This camera would capture images of the roadside assets and carriageway to the left of the vehicle. (Driving each road in both directions means video would be collected on both sides of the carriageway.) Experimentation with a camera and wide angle lens mounted on a test vehicle determined the optimal mounting specifications for the side camera: to be mounted 2.3m above the ground, mounted on the far left hand side of the vehicle, facing 60° to the left of the direction of travel, and facing down at the ground at an angle of 4° below the horizontal (See Figure 39 and Figure 40). This camera is known as the “side camera”. The side camera could be turned on or off during surveys but the procedure for standard GIPSICAM surveys would be to have the side camera turned on.
Consultation with internal clients and users also revealed a number of other shortcomings of the previous generation camera technology. An example was that the small writing on some traffic signs, such as the hours on a no parking sign, was sometimes hard to read from the front camera video. Video of the road shoulder, cracks in the pavement, and the guard fencing on the right hand side of divided carriageways are some other examples. The issue, however, was that different camera lens, settings and mounting specifications would be needed to capture each of the different video applications and requirements. It was unrealistic to install a large number of cameras for
66 Vehicle Selection and Modification
every possible application, so initially it was decided to install two additional cameras that would be preset to the two most useful applications.
Thus it was decided that a 3rd camera would be incorporated into the design of the new GIPSICAM vehicle. The purpose of this camera was to capture images of the traffic signs on the left hand side of the road. Experimentation with a camera and a telephoto lens mounted on a test vehicle determined the optimal mounting specifications for the 3rd camera: to be mounted 2.3m above the ground, mounted on the left of the front camera, facing 30° to the left of the direction of travel, and facing horizontal to the ground (See Figure 39 and Figure 40). This camera is known as “camera 3”. For standard GIPSICAM surveys camera 3 would be turned off.
It was also decided that a 4th camera would be incorporated into the design of the new GIPSICAM vehicle. The purpose of this camera was to capture images of the road shoulder on the left hand side of the road. Experimentation with a camera and a telephoto lens mounted on a test vehicle determined the optimal mounting specifications for the 4th camera: to be mounted 2.3m above the ground, mounted on the right of the front camera, facing 45° to the left of the direction of travel, and facing down at the ground at an angle of 30° below the horizontal (See Figure 39 and Figure 40). This camera is known as “camera 4”. The camera could be turned on or off during surveys but the procedure for standard GIPSICAM surveys would be to have camera 4 turned off.
The cameras needed to be mounted at the front of the vehicle so as to ensure the best, unobstructed view of the road in front of the vehicle and the road shoulder to the left of the vehicle. The concept of a camera turret evolved where a moulded turret would be built and incorporated into the front roof area of the vehicle. The turret would be high enough to be able to position all four cameras at a height of 2.3m above the ground. The turret would also have apertures for the cameras to “see out” of the vehicle from behind panes of glass.
67 Vehicle Selection and Modification
Figure 39. GCv3 standard camera angles (horizontal plane relative to direction of travel)
Upon deciding on the Mercedes Sprinter as the vehicle to be used, the concept for the camera turret changed slightly. The Sprinter had a high roof, which meant that the cameras could be positioned at 2.3m above the ground and still be lower than the roofline of the vehicle. This meant the camera turret did not need to protrude above the roofline, so the plans for the turret changed to incorporate the “body lines” of the Sprinter. This change also meant that the aerodynamics of the vehicle would not be altered as would have been the case if the turret was to protrude higher than the vehicle roofline, resulting in increased drag, increased fuel consumption, and impacting the quality of the collected road geometry due to the altered handling of the vehicle.
The initial prototype for the camera turret was constructed from wood, with PVC pipe used as camera barrels to allow the cameras to “see out” of the vehicle. The camera barrels were mounted on the turret and aligned according to the mounting specifications of each of the four cameras (See Figure 41).
68 Vehicle Selection and Modification
Figure 40. GCv3 standard camera angles (vertical plane relative to direction of travel)
However the lens angle of view, the length of the barrel and the diameter of the barrel were all dependent on each other. If one of the three changes, then one or both of the other two had to change accordingly. If the combination of lens angle of view, length of the barrel and diameter of the barrel were incorrect then the video captured from that camera would have “black ears” where the four corners of the video frame were capturing the inside of the barrel itself.
For example, a wide-angle lens required either a short length barrel or a large diameter barrel. If the length of the barrel was extended then the diameter of the barrel had to be increased. Accordingly, if the diameter of the barrel was to be decreased then the length of the barrel had to be shortened. Changing to a telephoto lens meant a longer barrel length and smaller barrel diameter combination could be used.
Given that each camera lens angle of view was fixed, and the mounting specifications such as height, position and orientation were known, then the corresponding barrel
69 Vehicle Selection and Modification
length and barrel diameter required, when installed relative to the existing bodylines of the vehicle, could be determined.
Figure 41. The initial design for the camera turret
An issue that became more prominent after building the model turret was “how do I implement the installation of the glass to keep the environment outside and away from the cameras”. The straightforward solution was to install the glass at the end of each barrel. The idea was sound, but there was a potential problem. The issue was that in order to clean the glass the vehicle staff would have to exit the vehicle and clean the glass from the front or side of the vehicle. Another idea to resolve this issue of cleaning the glass from within the vehicle, hence obviating the need for staff to exit the vehicle to clean the glass. Thus the idea of the “drop down” glass was explored, that is, the glass from each barrel could be removed while inside the vehicle, cleaned, and returned to position in the barrel. The problem with this idea though was that the position of the glass in the barrel would need to be far enough back that it could be accessed from
70 Vehicle Selection and Modification
within the vehicle, which was causing more issues relating to the length and diameter of the barrel and the problem with the “black ears” in the video frame.
It was at this point that the idea of a rectangular barrel instead of a round barrel was investigated. Rectangular barrels were made from cardboard for testing purposes (See Figure 42). This solved the issue of the “black ears” in the video frame. The rectangular barrels idea also seemed well suited to accommodate removable glass.
Figure 42. The rectangular barrel concept
Another simple idea was to have glass that could be fixed in place and cleaned using windscreen wipers. This idea was initially abandoned when the suggestion was put forward that the glass in front of the cameras needed to be perpendicular to the direction of the camera otherwise refraction would cause the images to be distorted, affecting image quality and asset capture accuracy. But would refraction of light through glass that was not perpendicular to the camera direction really cause image distortions? The angle of the light passing through the glass to each side of the camera lens would be
71 Vehicle Selection and Modification
slightly different, but once the camera was calibrated then the calibration matrix would account for any minor distortions. Thus the issue of having the glass perpendicular to the direction of the camera was found to not be an issue at all, and the idea of a single piece of glass was revisited. After some further testing it was decided to move the barrel idea back to the idea of a single large piece of glass, and the result was a turret with a large open aperture at the front for three forward facing cameras and a small aperture on the left side for a side facing camera. The front camera could be positioned to capture the video to the front of the vehicle and along the road, but mounting characteristics could still be modified at any time in the future if it was ever needed. The side camera could be positioned to capture the video to the left of the vehicle, but the mounting characteristics could still be modified at any time if needed. Finally, camera 3 and camera 4 could have any mounting characteristic required to suit the application. Cameras 3 and 4 could be pointed at the road pavement to video cracks, pointed to the right to collect video of roadside assets on divided carriageways, positioned to video traffic signs, positioned to video the road shoulder, or even positioned as a stereo pair for 3D applications. It was decided that the side camera barrel would be abandoned in favour of a small fixed side window with a sunshade.
A new prototype turret was constructed from cardboard and modification of the vehicle commenced. Continual testing was undertaken at every stage of the modification (See Figure 43). Initially a temporary camera shelf was installed inside the vehicle to assist with testing, before finally a permanent camera shelf was constructed and installed. The camera shelf was lined with vinyl to provide a professional finish, and black curtains were installed to stop any light from the rear of the vehicle reflecting off the front or side glass panes and causing artefacts in the video collected. The slot for the removable front glass pane was also constructed and implemented. Finally, a turret lid was added so that the front aperture of the turret could be protected when not conducting surveys.
The specifications for the front glass pane were provided to a specialist glass manufacturer and four identical moulded glass panes were produced. The extra three panes were to be replacement glass panes, if ever required. A wooden storage compartment for the spare front glass panes was constructed and installed in the vehicle cargo area, just behind the cargo barrier.
72 Vehicle Selection and Modification
Figure 43. Testing the new turret design
To assist with manual configuration of camera mounting specifications for different applications, camera mounts were developed with predefined camera angle settings. Four sets of wooden camera blocks were precisely manufactured at vertical angles from level to 30 degrees downwards from level, in increments of 1 degree. Then four sets of metal mounting bases that attach to the camera shelf were manufactured. Each metal mounting base had precisely measured guide holes drilled in them corresponding to straight ahead and parallel to the direction of travel, and additional sets of guide holes drilled at 15 degrees, 30 degrees, 45 degrees and 60 degrees each side of straight ahead. Thus using the combination of a camera block to set the vertical camera angle and a metal mounting base to set the horizontal camera angle, a range of camera mounting orientations can be quickly and accurately set by the vehicle staff corresponding to the specific survey application required. The metal mounting bases were securely fixed to the camera shelf for each of the camera locations, three at the front and one on the left.
73 Vehicle Selection and Modification
To install the cameras in the vehicle, first a metal base was securely attached to each camera. The metal camera base was then secured to an accurately manufactured block of wood, cut at a precise angle, corresponding to the vertical angle that the camera was to be set at. Next the wooden block was attached to a rubber block to provide vibration proofing, and securely attached to a metal base with guide pins. The guide pins of the lower metal camera base were inserted into the precisely measured guide holes of the camera shelf mounting base plate corresponding to the horizontal angle that the camera was to be set at. Finally the two bottom metal base plates were securely fastened together.
Figure 44. The finished designs for the camera turret
3.3.7 Computer/equipment rack
Equipment such as GPS receivers, computers and external hard drives needed to be securely housed in the cargo area of the vehicle. A search of the Internet revealed limited options for procuring a mass manufactured computer/equipment rack that suited the specialised needs of this project, hence the decision was made to have a custom designed computer/equipment rack built and installed in the vehicle. A rough design for the computer/equipment rack was taken to an industrial design and engineering company in Sydney, who undertook the detailed design and construction of the component. The primary features of the computer/equipment rack are its metal construction, powder-coated finish, two rigid shelves and two vibration-proofed shelves (See Figure 45).
74 Vehicle Selection and Modification
The rack is fixed to the floor of the vehicle cargo area, via rigid mounts, and fixed to the roof of the vehicle, via vibration-proofed mounts, to provide stability while also minimising vibration that may be propagated through the shell of the vehicle.
Figure 45. The computer/equipment rack
75 Vehicle Selection and Modification
3.3.8 Equipment operator work area
When GCv3 became operational the plan was that the equipment operator would conduct a road survey while sitting in the passenger seat of the vehicle. To control the survey progress and monitor the equipment status, the equipment operator required a work area to be constructed around the passenger seat. After careful consideration it was decided that the main components of the equipment operator work area would consist of a console with a display and input devices, communications equipment, and a stationary compartment to store printed materials and reference guides.
The initial idea for the implementation of the console was to remove the vehicle glove compartment and modify the vehicle dashboard area around the glove compartment, to create an area where a monitor, keyboard and mouse could be installed. The monitor would be installed in the cavity created by the removal of the glove compartment, and positioned close to the windscreen and vehicle firewall, with some ability for the monitor to be tilted up or down and left or right. A flat area in front of the monitor would allow the permanent mounting of a PC keyboard. Flat areas on either side of the keyboard would allow the use of a PC mouse and provide a space for placing reference materials such as maps.
However, in May 2005, a couple of months after ordering the vehicle, it was discovered that the vehicle was to be manufactured with the passenger airbag fitted. Hence other console had to be considered. For example, a touch screen monitor mounted on an arm at the centre of the dash, away from both passenger and driver airbags, or mounted to the floor of the vehicle, in between the passenger and driver seats. However such an option was ruled out as it uncomfortable to “reach out” to the monitor all day long, which raised a possible OH&S issue. In addition the touch screen accumulated fingerprints rapidly.
A standard keyboard could be placed on the equipment operator’s lap, but it is not particularly OH&S friendly and where would the mouse go? The answer was not to use one. Instead, a computer monitor would be mounted on an arm, located in the centre of
76 Vehicle Selection and Modification
the dash or between the seats. Input to the console would be provided by a flexible USB keyboard, which would present less risk of injury in the case of an accident than a standard hard plastic keyboard. A “stable table” would be utilised by the equipment operator as a flat work area to place the flexible keyboard, and to place reference materials such as maps or for completing operational forms such as the vehicle run sheets. Finally, a trackball mouse would be mounted within comfortable reach of the equipment operator and in-between the seats. The computer monitor selected was a DELL UltraSharp 15 inch LCD monitor, 1505FP, lightweight and compact, providing a clear image at a resolution of 1024 x 769 @ 75 Hz. It has both a DVI-D (digital) and a VGA (analog) connector for connecting to a computer video display port. It also has 4 USB 2.0 ports for connecting devices such as a flexible keyboard, a trackball mouse, a memory stick, etc.
Figure 46. The monitor arm and Dell monitor
A quick search of the Internet revealed limited options for procuring a mass manufactured monitor arm that suited the specialised needs of this project, hence it was
77 Vehicle Selection and Modification
decided to have a custom designed monitor arm built and installed in the vehicle. A rough design for the monitor arm was taken to an industrial design and engineering company in Sydney, who undertook the detailed design and construction of the component. The primary characteristics of the monitor arm are its metal construction, powder-coated finish, breakaway design for safety, and two-way adjustment capability to tilt the monitor up or down and left or right. The monitor arm was fixed to the floor of the vehicle in-between the seats via a rigid mount. The monitor was attached to the mounting-plate at the other end of the monitor arm (See Figure 46).
Two flexible keyboards were procured from local computer shops. Their features include that they were waterproof, easily cleaned, had USB connection, lightweight and cheap to replace.
A set of “stable tables” was procured from a local shopping centre.
Figure 47. The equipment operator work area (the passenger side)
78 Vehicle Selection and Modification
After researching the various trackball mice available commercially, the Logitech Marble Mouse was selected. The features of this mouse include large trackball, large buttons, optical precision, USB connectivity, and ergonomic shape. It was easy to use and enabled precise control of the cursor on the screen. The mouse was mounted on a platform forming part of the wooden stationary compartment located in-between the seats (See Figure 47).
The stationary compartment was made from wood and designed to fit in the space between the driver and passenger seats. It had a number of features including a mounting platform for the trackball mouse, a mounting platform for the vehicle mobile phone hands-free cradle, a mounting platform for the recon data collection device, storage area for water and drinks, storage area for stationary, and storage areas for printed materials such as road maps, survey plans and run sheets.
3.3.9 GIPSITRAC odometer
The decision had been made to utilise a GIPSITRAC box in the new vehicle to collect road geometry and inertial data. Synchronisation of the data collected by the GIPSITRAC box is accomplished via the use of a GPS and an odometer. A GPS is built-into the GIPSITRAC box, however the GIPSITRAC requires the use of an external sensor to generate the odometer pulses. A number of pulse generating technologies and implementations were investigated, such as the PCA INSG 10HR A2A1/0020 Incremental Shaft Encoder (58mm diameter) mounted on a rear wheel and the Mercedes Sprinter ABS system providing pulses from the rear wheels, but the technology selected to provide the odometer pulses was the Phillips Semiconductors KMI 15/1 Integrated Rotational Speed Sensor.
The KMI 15/1 was implemented using the drive shaft of the vehicle, rather than attaching it to one of the rear wheels. A steel cog with 40 teeth was manufactured and inserted into the vehicle drive shaft assembly. A sensor mount was manufactured and the sensor was attached to the sensor mount. Next, the sensor mount was attached to the vehicle differential housing and positioned so that the sensor was aligned with the teeth
79 Vehicle Selection and Modification
of the steel cog attached to the drive shaft. Finally, the sensor was carefully soldered to a cable to enable connection to the GIPSITRAC box (See Figure 48).
Figure 48. The steel cog and KMI 15/1 rotational speed sensor
3.3.10 GPS antenna mount points
Even though the decision was made to place all equipment inside the vehicle where the environment could be controlled, there still was a requirement to place some equipment outside the vehicle. This is the case of the GPS antennas which need an unobstructed view of the sky to receive direct satellite signals. There was to be three GPS receivers utilised in the new vehicle: a primary GPS, a back-up GPS, and a GPS used for timing.
The preferred solution was to not place a rack or any other “high” platform on the roof of the vehicle to mount the GPS antennas on, as this would produce air resistance when driving which would lead to higher fuel consumption, unnecessary noise, and
80 Vehicle Selection and Modification
potentially cause vehicle stability issues in strong winds that could affect the accuracy of the road geometry collected. The mobile phone and CB radio antennas were to be mounted above the cab of the vehicle and the warning lighting was to be mounted on the roof at the rear of the vehicle. To minimise the chances of interference of the GPS from the mobile phone and CB radio antennas placing the GPS antennas somewhere in the middle of the vehicle seemed the best option. Inspecting the roof of the vehicle revealed a “rib” pattern of protrusions in the roof, which could potentially limit the mount locations of the antennas.
Close inspection of the roof revealed three slightly elevated mounds, one which was circular in shape and located on the centreline of the roof, and the other two which were rectangular in shape and located behind the circular mound and each equally offset from the centreline of the roof. These three mounds initially appeared to be good candidates for mounting the three GPS antennas.
However, the mounts were very close together and could potentially create a “canyoning” effect on each other by obscuring satellites from each antenna. Thus it was decided to only use the circular mound located on the centreline of the roof and position the other two antennas alongside the circular mound, which was an equal distance from the “origin” of the vehicle in the longitudinal direction as the primary GPS antenna, but equally offset towards each edge of the roof in the transverse direction. Spacing the three antennas apart should minimise any potential “canyoning” effect. The next decision was to decide which antenna should be placed where. The primary GPS is the most important as it provides the accurate real time positioning information. Thus the primary GPS antenna mounting bolt was installed in the centre of the raised circular mound in the middle of the roof and on the centreline of the roof.
The secondary GPS, also referred to as the back-up GPS, is the next most important of the three GPS. One possible reason for the primary GPS to fail could be that the vehicle is in an enclosed environment such as the CBD, which is another example of “canyoning”. However, in such environments it is usually the case that the road currently being surveyed is a two-lane road, thus moving the GPS antenna further to the right of the vehicle may result in a larger view of the sky or a view of a different part of
81 Vehicle Selection and Modification
the sky where more satellites are present, and thus being able to calculate a position when the primary GPS may have failed to do so. Thus the secondary GPS antenna mounting bolt was installed an equal distance from the origin of the vehicle in the longitudinal direction as the primary GPS antenna, but offset to the far right of the roof from the centreline in the transverse direction.
Figure 49. GPS antenna mount points
The GIPSITRAC GPS is used for timing and thus can operate with a lower number of visible satellites than the other two GPS. It was decided that it could be mounted on the left side of the roof, an equal distance from the origin of the vehicle in the longitudinal direction as the primary GPS antenna, but offset to the far left of the roof from the centreline in the transverse direction.
With the GPS antennas located on the roof of the vehicle, a way of feeding the antenna cables back inside the vehicle was needed. The solution was a small waterproof aperture in the roof of the vehicle located along the vehicle centreline and positioned back from
82 Vehicle Selection and Modification
the centre GPS antenna mount point, which would allow GPS antenna cables to pass in/out of the vehicle (See Figure 49).
3.3.11 Survey/warning lights
Regulatory requirements dictate that amber warning lights are to be mounted on the vehicle and utilised when surveys are in operation. A Hella K-LED 1608AMB4A Amber LED Rotating Strobe was installed on the roof of the vehicle at the rear of the cargo area. Two Hella 95903710 Amber LED high intensity surface mount flashing lights were installed on the rear doors of the vehicle near the roof, and positioned each side of the rotating strobe. A Hella 4442 Amber Illuminated On/Off switch (12V) installed on the dash is used to operate both the amber flashing lights and the amber rotating strobe.
Two Hella 95903700 Red LED high intensity surface mount flashing lights were installed on the rear doors of the vehicle as a safety initiative. The red flashing lights can be turned on to signal vehicles behind that the GIPSICAM vehicle may be slowing down or pulling over to the side of the road. A Hella 4474 Red Illuminated On/Off switch (12V) installed on the dash is used to operate the red flashing lights.
Two Hella Comet 450 series driving lights were also fitted to the vehicle as a safety precaution, in case the vehicle is driven at night on country roads.
3.3.12 Magnetic signage, stickers and RTA decals
When the vehicle is surveying roads it is likely that the vehicle will operate in a different manner to other road users. For example, the maximum survey speed is 80km/h so if surveying a road that has a posted speed limit of 100km/h then it is likely that other motorists will catch up and overtake the vehicle. Thus a way was needed to inform the other motorists of the different vehicle behaviours. Magnetic signage, stickers and RTA decals were produced and attached to the vehicle to warn and inform other motorists of the possibility of unexpected driver behaviour when the vehicle was
83 Vehicle Selection and Modification
surveying. RTA decals were attached to the vehicle on the top left and top right of the bonnet, the top left and top right of the rear doors, the top centre of the driver’s door and the top centre of the passenger’s door.
Figure 50. GCv3 with magnetic signage (front view)
Reflective red and yellow warning stripes (left and right) were affixed to the vehicle’s waistline, from the passenger/driver door to the rear of the vehicle and across the full length of the two rear doors (See Figure 50).
Removable magnetic signs were manufactured to inform other motorists of the operation of the vehicle. Magnetic signs strategically positioned on the bonnet and sides of the vehicle contain the following wording in black text on a non-reflective yellow background: “Road Survey Vehicle”, “GIPSICAM”, ““Caution – Survey Vehicle”, and Video Capture in Progress”. Magnetic signs strategically positioned on the rear doors of the vehicle contain the following wording in black text on a reflective yellow background: “Van May Operate Below Speed Limit” and “Pass when Safe”. An
84 Vehicle Selection and Modification
additional magnetic sign strategically positioned on the rear doors of the vehicle contains the following wording in black text on a non-reflective orange background: “Caution Van May Slow” (See Figure 51).
Figure 51. GCv3 with magnetic signage (rear view)
3.3.13 General/minor modifications
A number of other minor modifications were undertaken on the vehicle: Hayman Reese R1603 tow hitch with step bar. 2 x mounting brackets and sunshades to suit Elf Mk5 & Elf Mk6 Odometers, installed on the dash. 2 x mobile phone hands-free cradles installed on the mounting platform provided by the wooden stationary compartment within the equipment operator work area. UHF CB radio.
85 Vehicle Selection and Modification
Install mobile phone and CB radio antennas on the roof, driver’s side of vehicle, towards the front of the vehicle. Convex mirrors to be mounted on passenger’s door, driver’s door and rear doors. Reversing alarm. Window tinting on drivers door, passengers door, and rear cargo doors. Tinted strip across windscreen. Window weather shields for driver’s door and passenger’s door. Cabling and cable channels around the vehicle. CO2 fire extinguisher mounted in cargo area of vehicle. Dry chemical fire extinguisher mounted beside driver’s seat in front cab of vehicle. 12V lights in the cargo area of vehicle.
3.4 Concluding remarks
The GCv3 vehicle has now been operational for five seasons. The selection of the Mercedes Sprinter 316 CDI for the GCv3 vehicle was an excellent choice, having performed very reliably over these years. The GCv3 vehicle modifications in general have also performed exceptionally and have help realise the three primary design considerations of the GCv3 vehicle, which were: quality/accuracy, efficiency and OH&S. Overall, the design and implementation of the various vehicle modifications were very good and would be recommended for consideration/incorporation into other MMS vehicles being developed in the future, such as GCv4, with the only exceptions listed below.
With hindsight, after operating the vehicle for five seasons, it was found that the design and implementation of the air conditioning was adequate, though still lacking (See section 3.3.3 Air conditioning). The OEM compressor struggles a little on really hot days, greater than 40 degrees Celsius, to feed cold air to both air conditioners. Another issue is that when operating on dirt roads, such as the northern part of the Silver City
86 Vehicle Selection and Modification
highway in the far west of the state, fine dust is sucked through the air conditioning filters and out the vents to be deposited on the equipment within the cargo area.
The design and implementation of the equipment electrical circuit were very good, with two exceptions, which were the mounting of the 2nd alternator and the wiring of the distribution switch (See section 3.3.5 Equipment electrical system). A few months after taking delivery of the vehicle from the body builder, who installed the 2nd alternator, during road tests it was found that the fan belt for the 2nd alternator had been shredded which meant that the alternator was not delivering charge to the equipment batteries anymore and thus the batteries had run down. Taking the vehicle back to the body builder revealed that the 2nd alternator pulley was misaligned with the crankshaft pulley. The 2nd alternator was realigned and a new fan belt installed.
However, six months later the same voltage loss problem occurred. This time it was the tensioning bracket that broke. A reinforced tensioning bracket was installed. Eighteen months later a rattling noise was heard under the bonnet of the vehicle, which revealed that one of the alternator mounts had broken off. The mount was reinforced and off the vehicle went again. Six months later the tensioning bracket broke again. Then the 2nd alternator failed. There are still issues relating to both the alignment of the 2nd alternator and also the securing of the 2nd alternator resulting in vibration, which in combination with the high forces applied via the tensioned fan belt, is resulting in the “current weakest mount point” of the 2nd alternator breaking. A number of body builders and auto electricians have tried to resolve the issue but it is an outstanding problem to this day. Two possible solutions for this problem are rethinking the current mounting configuration or find an alternative mounting location for the 2nd alternator that is not affixed to the OEM alternator.
The other issue relates to the understanding of the wiring of the distribution switch. The body builder had not provided a wiring diagram showing the “as built” wiring so it was assumed that the wiring was built as requested, with the “Both” setting wired for the OEM and 2nd alternators to charge the vehicle circuit and the equipment circuit respectively. A few months after delivery of the vehicle from the body builder at the time of the first alternator problem, a test of the distribution switch revealed that it was
87 Vehicle Selection and Modification
not behaving as expected. After a closer inspection of the implemented wiring and further experimentation it was determined that the distribution switch was wired to give the following selectable options: Off – The OEM and 2nd alternators charged the vehicle circuit and the equipment circuit respectively. Both – Both alternators charged both the vehicle circuit and the equipment circuit. 1 – The OEM alternator charged both the vehicle circuit and the equipment circuit. 2 – The 2nd alternator charged both the vehicle circuit and the equipment circuit.
With hindsight, after operating the vehicle for five seasons, it was found that the design and implementation of the camera turret overall was very good, with the one exception being the removable glass (See section 3.3.6 Camera turret).
The idea of the removable glass was very alluring in terms of the vehicle staff not having to exit the vehicle to clean the front glass, but the reality was that it was a failure. When first constructed, the waterproofing of the area around the glass was not very good and if operating the vehicle in rain then water would run down the glass, collect in the guide slots and then finally drip inside the vehicle. The solution of adding waterproofing around the guide slot resulted in splattered grasshoppers being wiped off the glass and onto the waterproofing when the glass was removed and smudged back on again when the glass was reinserted. In addition, if it was raining outside then the glass could not be removed, as the rain would then splatter onto the camera lens. Worse still, if it was raining and windy then rain could get blown on the cameras themselves which was not a good outcome. The glass could not be removed in dusty areas either. So effectively the only real use of the removable glass was when operating in urban conditions where insects were not an issue, when it was not raining and not dusty. Otherwise the vehicle staff had to exit the vehicle to clean the front glass from the passenger side of the vehicle using a squeegee.
In hindsight a similar concept to the camera turret would be implemented but the removable front glass pane would be replaced by a fixed glass pane with windscreen
88 Vehicle Selection and Modification
wipers that covered the entire area of the glass pane, and which had a control mechanism to allow manual operation or variable speed automated operation.
See section 9.2 Future research and development opportunities for GIPSICAM.
89 Equipment Selection and Systems Integration
CHAPTER 4 EQUIPMENT SELECTION AND SYSTEMS INTEGRATION
4.1 Introduction
This chapter describes the processes and actions taken in the review and selection of equipment, and the integration of the hardware intended to produce the GIPSICAM data.
4.2 Equipment selection
The data collection equipment within the GCv2 system consisted of procured items that were essentially modular in construction (Entriken, 2005). That is, the different components were connected to each other via standard connections and communicated via standard protocols. This made it relatively easy to change equipment if required due to an upgrade or the replacement of a faulty item. The primary data collection equipment within the GCv2 system is listed in chapter 1 (See 1.5 A brief history and overview of GIPSICAM v2 (GCv2)).
4.2.1 DGPS receiver with real-time differential corrections
A fundamental requirement for a MMS is to know spatially where it is at any point in time while surveying so that it can relate the image (or other) data collected to a unique position on the Earth. Previous experience with GCv2 had shown that utilising a real time differentially corrected GPS solution was a very effective way to determine absolute positioning of the vehicle when surveying. The existing primary GPS technology utilised in GCv2 was a Trimble Pro XRS DGPS.
The Trimble Pro XRS DGPS has the capability to receive and apply real time OmniSTAR differential corrections via the OmniSTAR VBS service. The Trimble Pro
90 Equipment Selection and Systems Integration
XRS DGPS was still being supported by Trimble in 2005, with firmware upgrades available to ensure the GPS was operating correctly. The decision was made to utilise the Trimble Pro XRS DGPS with the OmniSTAR VBS service within GCv3 as the primary positioning equipment (See Figure 52 and Figure 53).
In mid-2007, after the vehicle had been operating for 12 months, the newly released Trimble Pro XRT DGPS was procured as an upgrade to the existing technology. The advantage of the XRT over the XRS was that the XRT can receive corrections from the OmniSTAR HP service, promising an accuracy of 10cm (2D). In addition, the XRT had an option to receive GLONASS signals and use them in calculating positions. Finally, the XRT had a redesigned antenna that claimed to provide optimal performance of the XRT.
Figure 52 Trimble Pro XRS, rear view
91 Equipment Selection and Systems Integration
Figure 53 Trimble Pro XRS specification (Trimble documentation, modified)
92 Equipment Selection and Systems Integration
4.2.2 GPS receiver for post-processed differential corrections
Reliability was an important consideration for GCv3. Having a redundant secondary GPS receiver that can collect absolute positions in case there is a problem with the primary GPS receiver means reduced risk of a failed survey. However, is there a risk if using the same real time differential correction service with both the primary and the secondary GPS? Where is the redundancy for the differential corrections? It was decided that utilising a secondary GPS with post-processed differential corrections would mitigate any risk associated with problems related to the OmniSTAR VBS service.
The convenience of the OmniSTAR VBS service is that the differential corrections are received in real time, anywhere in NSW. However, in the unlikely situation that the secondary rover GPS data would be needed the strategy was to post-process the differential corrections back in the office using base station data. The RTA did operate a GPS Base Station at the Rockdale works office, where GIPSICAM was based, at the time of development of GCv3. The base station consisted of a Trimble Pro XL (12 channel) GPS receiver and the Pathfinder Community Base Station (PFCBS) software operated 24/7 collecting base station data.
The School of Surveying and Spatial Information Systems at UNSW also operated a GPS base station at their Kensington campus and provided the data free online. Finally, SydNET and subsequently the CORSnet-NSW network came online providing data from base stations initially located in the Sydney basin, and then across all of NSW.
The existing secondary GPS technology utilised in GCv2 was a Trimble Pro XL (8 channel) GPS.
Trimble no longer supported the Trimble Pro XL GPS in 2005, however some firmware upgrades were available and testing revealed that the GPS was operating satisfactorily.
93 Equipment Selection and Systems Integration
The decision was made to utilise the Trimble Pro XL GPS within GCv3 as the secondary absolute positioning equipment (See Figure 54). The Trimble Pro XL GPS would operate as a rover in the GCv3 vehicle and the GPS data could be post-processed to differentially correct for errors, if and when it was required.
Figure 54 Trimble Pro XL specification (Trimble documentation, modified)
4.2.3 ARRB GIPSITRAC
GPS can provide accurate absolute positions, but it suffers from multipath, canyoning effects, and signal loss under trees and in tunnels. An inertial navigation system (INS) on the other hand provides continuous (relative) positions and operates when GPS does not. Thus, as has been the case for many applications over the last two decades, combining GPS and INS enables a more robust vehicle positioning solution.
The GIPSITRAC was utilised in GCv2 to provide relative positioning from two micro- electromechanical system (MEMS) gyroscopes and a rotational speed sensor.
94 Equipment Selection and Systems Integration
GIPSITRAC also provided road geometry data collected using MEMS accelerometers, MEMS gyroscopes and a rotational speed sensor. In addition GIPSITRAC provided synchronisation mechanisms using an autonomous GPS as an accurate clock and the rotational speed sensor as an odometer. All of these capabilities were required in GCv3. The only capability not required from GIPSITRAC in GCv2 was the video time-code interface as the decision had been made to record digital video direct to hard drives via software on the vehicle computer.
However, the GIPSITRAC from GCv2 was not utilised in GCv3. The primary reason for using a different GIPSITRAC was so development and testing of GCv3 could occur concurrently while GCv2 was out surveying the GIPSICAM 2005/2006 survey season. The assumption was that both GIPSITRAC devices would be identical, however that was not the case.
The procured GIPSITRAC was originally installed in a station wagon car and was used during the 1990’s to collect road geometry data. However this dedicated function was no longer required as GIPSICAM also provided road geometry data. The procured GIPSITRAC did not have the video time-code interface that the GIPSITRAC in GCv2 had, as it was no longer required. Another difference between the two GIPSITRAC devices was that they were designed to be orientated differently when mounted in a vehicle. Following directions provided by the Australian Roads Research Board (ARRB), a minor modification was made to the new GIPSITRAC to change the orientation of the accelerometers, so that the procured GIPSITRAC could be installed in GCv3 in a longitudinal orientation rather than the transverse orientation it was originally designed for.
The GIPSITRAC, providing synchronisation, relative positioning and vehicle attitude, was comprised of: Motorola PVT6 GPS receiver, used for GIPSITRAC sensor synchronisation. Two Murata ENV05S MEMS gyroscopes, used to determine the direction of travel and the horizontal radius of the vehicle path. Two Honeywell QA700 accelerometers, used to determine the grade, cross-fall and vertical radius of the vehicle path.
95 Equipment Selection and Systems Integration
Philips Semiconductors KMI 15/1 rotational speed sensor, used for GIPSITRAC sensor synchronisation and also to provide data on the distance travelled. Microprocessor, used to control the GIPSITRAC sensors and all I/O functions.
The GIPSITRAC is securely mounted in the centre of the vehicle and over the rear differential, which is the origin of the vehicle reference frame (See section 3.3.1 Equipment/storage cabinet).
Figure 55. ARRB GIPSITRAC INS (with the top removed for maintenance)
4.2.3.1 Motorola PVT6 GPS
The GIPSITRAC contains a Motorola PVT6 GPS, a 6 channel parallel L1/CA code receiver (ARRB, 1995), utilised as an accurate clock for synchronisation of the GIPSITRAC sensors.
96 Equipment Selection and Systems Integration
Figure 56 Motorola PVT6 specification (Motorola documentation, modified)
4.2.3.2 Murata ENV05S MEMS gyroscopes
The GIPSITRAC contains two Murata ENV05S gyroscopes (ARRB, 1995), which are used to determine the direction of travel and the horizontal radius of the vehicle path. The Murata ENV05S is a piezoelectric vibrating gyroscope.
Figure 57 Murata ENV specifications (Avnet Kopp, 2003)
97 Equipment Selection and Systems Integration
4.2.3.3 Honeywell QA700 accelerometers
The GIPSITRAC contains two Honeywell QA700 accelerometers (ARRB, 1995), one placed in a longitudinal orientation and the other in a transverse orientation, which are used to determine the grade, cross-fall and vertical radius of the vehicle path.
Figure 58 Honeywell QA700 specifications (Honeywell, 2004 and modified)
4.2.3.4 Philips Semiconductors KMI 15/1 rotational speed sensor
The Philips Semiconductors KMI 15/1 rotational speed sensor is used for GIPSITRAC sensor synchronisation and to provide data on the distance travelled. The KMI 15/1 is
98 Equipment Selection and Systems Integration
used in conjunction with a manufactured 40-toothed steel cog that is fitted in-line with the drive shaft (See section 3.3.9 GIPSITRAC odometer).
4.2.4 FOG Gyroscope
The GIPSITRAC contained two Murata ENV05S gyroscopes that output relative positioning information. This MEMS technology required daily calibration, which can affect the accuracy of the output turn-rate data. Thus there had to be procedures in place to ensure calibration was performed correctly (See section 6.2 Calibration procedures). In addition, the output data accuracy of these gyroscopes is dependent on the temperature of the MEMS sensors during operation (ARRB, 1995). Once again, procedures had to be put in place to ensure that the GIPSITRAC was ‘warmed up’ to operating temperature before commencing the daily calibrations. Finally, while the MEMS gyroscopes did provide reliable output turn-rate data, they had occasionally produced “obviously incorrect data” that suggested the vehicle was driving around in circles for the whole survey.
However, discovering there is a problem with the gyroscope data after the end of the survey was not going to fix the problem. The section of road with the problem data would have to be re-surveyed. There was a limited form of redundancy provided by having two MEMS gyroscopes in the GIPSITRAC, but if there was a calibration problem or a temperature problem then both MEMS gyroscopes could be affected. Thus a redundant gyroscope was needed that could be used to confirm the accuracy of the MEMS gyroscopes. The solution was to install an additional independent gyroscope, which would provide both redundancy and validation. The equipment that was found to provide the solution was a KVH DSP-3000 fibre optic gyroscope (FOG).
The KVH DSP-3000 FOG does not need to be calibrated by a user, thus removing the risk of calibration errors. The DSP-3000 utilises optical technology to determine changes in vehicle attitude, which is not as affected by fluctuations in temperature. Finally, the DSP-3000 is an independent piece of equipment which would provide an
99 Equipment Selection and Systems Integration
alternative data stream in the event the primary MEMS gyroscopes were not functioning correctly.
A KVH DSP-3000 FOG was procured and installed in the GCv3 vehicle cargo area (See section 3.3.1 Equipment/storage cabinet).
Figure 59. KVH DSP-3000 FOG
4.2.5 Vehicle odometers
A additional application of GIPSICAM was the accurate measurement of a section of road, such as was required for the Point-to-Point Speed Zone trials in 2004 (SMH, 2004). Thus the capability to accurately measure a section of road was included in the development of GCv3. The GIPSITRAC that was to be installed in the GCv3 vehicle utilised a KMI 15/1 integrated rotational speed sensor that performs the function of an odometer. A steel toothed cog mounted on the vehicle drive shaft was to be used to
100 Equipment Selection and Systems Integration
generate the distance pulses (See sections 3.3.9 GIPSITRAC odometer and 2.3.3.1 Variable Reluctance sensor).
GCv2 utilised Elf Mk5 and Mk6 odometers as survey tools, to assist the vehicle staff in determining how far it was to the end of the road section currently being surveyed. This was a manual process and one that was to be automated by software in GCv3, hence the Elf odometers were no longer needed. They could however be used as an additional independent odometer for applications involving the measurement of a section of road. Thus the decision was made to utilise the Elf odometers in the GCv3 vehicle as second independent odometers. The Elf odometers were connected to the vehicle electrical system and the vehicle ABS pulses were used to determine distance travelled.
Finally, a third independent odometer was sourced; it was the CORRSYS-DATRON CORREVIT LF II P optical sensor. This sensor was recommended by the RTA’s Camera Enforcement Branch (their title at the time of development) and is the same technology that is used by Formula 1 racing cars for performance testing.
4.2.6 Video equipment
The video equipment in the GCv2 vehicle was causing many problems. Recording onto a mini DV tape meant that a survey run had a defined maximum time limit. Every single survey run needed to be carefully planned to ensure that there would be enough time to get to the end of a section without running out of tape. Stopping too often is inefficient due to the time it takes to stop and restart, between 15-30 minutes depending on the level of experience of the vehicle crew. As an example, if using 90-minute tapes, you're the survey runs had to have at least 15 minutes spare, requiring them to be about 75 minutes long. Thus if surveying for 8 hours in a day, one would expect to spend between 1 to 2 hours of that day conducting end-of-run and start-of-run procedures, and not actually collecting survey data.
It was often the case that the video heads would get dirty and need to be cleaned. Occasionally the tape would break or get “chewed up”. A failed run would require a re-
101 Equipment Selection and Systems Integration
survey. But if a 1 hour run fails then up to 2 hours could be spent going back to the start and re-surveying it again. There was also the safety issue of the operator having to continually get out of the vehicle to go into the cargo area of GCv2 to change the tapes. Thus the decision was made to utilise video equipment that could record direct to a hard drive.
Another consideration was that the camera should have a resolution greater than PAL, which is what was available in GCv2.
4.2.6.1 Video Cameras
After some more research on the Internet and discussions with digital video camera suppliers, four camera bodies were evaluated in a trial: Sony DFW-SX910 (Sony, 2004) o 1/2” PS CCD (Sony ICX205AK) o C mount o 1280x960 @ 7.5fps, YUV 4:2:2 / Mono8 (Raw) Sony DFW-X710 (Sony, 2004) o 1/3” PS CCD (Sony ICX204AK) o C mount o 1024x768 @ 15fps, YUV 4:2:2 / Mono8 (Raw) Sony XCD-SX910CR (Sony, 2003) o 1/2” PS CCD (Sony ICX205AK) o C mount o 1280x960 @ 15fps, Mono8 (Raw) o 1280x950 @ 7.5fps, Mono16 (Raw) Basler A102FC (Basler, 2005) o 2/3” PS CCD (Sony ICX 285 ExView) o C mount o 1388x1038 @ 15fps, YUV 4:2:2
102 Equipment Selection and Systems Integration
Figure 60. Sony DFW-SX910 camera and Fujinon lens
Visual artefacts such as vertical smearing and blooming are common CCD sensor technology problems (Total Turnkey Solutions, 2005).
It was found that the Basler A102FC not suitable, as it did not have auto-exposure capabilities, considered essential as the vehicle would be operating in different lighting conditions, such as in direct sunlight on an open highway in western NSW, or in indirect light as for roads through one of the NSW state forests, or in artificial lighting as in the Sydney Harbour Tunnel. There was also the difficulty in acquiring an appropriate wide-angle lens to suit a 2/3” CCD camera.
The Sony DFW-X710 was very similar to the Sony DFW-SX910, but had a maximum resolution of only 1024x768, which was the lowest of all the cameras trialled, and not a mega-pixel CCD. It was ruled out for this reason.
103 Equipment Selection and Systems Integration
The Sony XCD-SX910CR streamed raw video, which meant real time processing or post-processing could be carried out. Real time processing would put a load on the vehicle systems so that was not an option. Post-processing was a possibility. The Sony XCD-SX910CR had a maximum resolution of 1280x960, however the Mono16 (Raw 10 bit) only had a maximum frame rate of 7.5fps, not the 15fps frame rate it had for Mono8 (Raw 8 bit).
The Sony DFW-SX910 supported a resolution of 1280x960 at 7.5fps, the same as the Sony XCD-SX910CR with Mono16, however the Sony DFW-SX910 also supported YUV 4:2:2, which in the end was the deciding factor in selecting the Sony DFW-SX910 as the camera to use in the GCv3 vehicle.
Four Sony DFW-SX910 cameras were procured and installed in the vehicle (See section 3.3.6 Camera turret).
4.2.6.2 Lens
The GCv3 vehicle camera system was to be modular and configurable so that GCv3 could be used for any number of applications. Accordingly, a range of lenses for the cameras to suit a wide range of possible applications was needed. After some initial research, and taking into account the cameras that were tested at the time, the specifications included: needed to be C mount, 1/2” format, mega-pixel lenses with a manual iris and a manual focus. A number of lenses that covered a range of horizontal angles of view from wide-angle to telephoto were needed. Five camera lenses were acquired for a trial: Fujinon DF6HA-1B (Fujinon, 2005 A) o 6mm, manual focus lockable, mega-pixel o F1.2 to F16, manual iris lockable o 1/2” or 1/3” or 1/4", C mount o 56.15° x 43.6° (HxV @ 1/2") o 0.1m to infinity focusing range Fujinon HF9HA-1B (Fujinon, 2005 B)
104 Equipment Selection and Systems Integration
o 9mm, manual focus lockable, mega-pixel o F1.4 to F16, manual iris lockable o 2/3” or 1/2” or 1/3”, C mount o 39.15° x 29.9° (HxV @ 1/2") o 0.1m to infinity focusing range Avenir HTCH0484 (Avenir, 2004 A) o 4.8mm, manual focus, mega-pixel o F1.4 to Close, manual iris o 1/2” or 1/3”, C mount o 81.4° x 69.4° x 54.2° (DxHxV @ 1/2") o 0.2m to infinity focusing range o -10°C to +50°C Avenir HZCH08551 (Avenir, 2004 B) o 8.5mm to 51mm, manual focus, mega-pixel o F1.2 to Close, manual iris o 1/2” or 1/3”, C mount o 50.7° x 41.2° x 31.3° (DxHxV @ 1/2" & 8.5mm) o 8.9° x 7.2° x 5.5° (DxHxV @ 1/2" & 51mm) o 1.1m to infinity focusing range o -10°C to +50°C Tamron 23FM25SP (Temron, 2004) o 25mm, manual focus lockable, mega-pixel o F1.4 to F22, manual iris lockable o 2/3” or 1/2” or 1/3”, C mount o 14.6° x 11.0° (HxV @ 1/2") o 0.15m to infinity focusing range o -10°C to +60°C
It was decided that the Avenir HZCH08551, the manual zoom lens, was unsuitable on the grounds that allowing the focal length to be manually set leads to possible issues relating to replication of a survey where the focal length was not recorded or was set incorrectly, or performing an unsuitable survey where the focal length was incorrectly set.
105 Equipment Selection and Systems Integration
It was found that the Avenir HTCH0484, with a 4.8mm focal length, produced some distortions around the edge of the image. This was to be expected from a wide-angle lens such as this.
The Tamron 23FM25SP, with a 25mm focal length, was found to be possibly suitable for capturing signs ahead of the vehicle. It was better to video a road sign further away from the vehicle when it was closer to perpendicular with the front of the vehicle, rather than trying to video the sign as it passed by to the left of the vehicle and was therefore visible only on an angle.
The Fujinon HF9HA-1B, with a 9mm focal length, was found to be possibly suitable for facing down at the pavement and videoing cracks. The lens had less distortion around the edges of the image but still provided good coverage of the road pavement.
The Fujinon DF6HA-1B, with a 6mm focal length, was found to be the most suitable of the lenses tested for the front camera and the side camera. It covered a large area of the front and the side of the vehicle, acquiring good quality images. Two Fujinon DF6HA- 1B lenses were used with the front and side cameras. Two Fujinon HF9HA-1B lenses were to be used for special purpose cameras. Two Tamron 23FM25SP lenses were also to be used for special purpose cameras. The Fujinon DF6HA-1B lenses were attached to the front and side cameras and used as the standard lenses for the GIPSICAM surveys.
However, after operating the GCv3 vehicle for its first season in 2006/2007 it was found that the 6mm Fujinon DF6HA-1B lenses on the front and side cameras were not providing the coverage that was wanted. It was decided that a lens with a slightly wider angle would be needed, however not one that would cause excessive distortions around the edge of the image. After some research on the Internet and discussions with digital video camera suppliers, the following camera lens was evaluated: Computar H0514-MP (Computar, 2006) o 5mm, manual focus lockable, mega-pixel o F1.4 to F16, manual iris lockable o 1/2” or 1/3” or 1/4", C mount
106 Equipment Selection and Systems Integration
o 76.7° x 65.5° x 51.4° (DxHxV @ 1/2") o 0.1m to infinity focusing range o -20°C to +50°C
The Computar H0514-MP, with a 5mm focal length, was found to be a suitable replacement for the Fujinon DF6HA-1B. The H0514-MP covered a horizontal angle of view of 65.5°, compared to the DF6HA-1B’s angle of view of 56.15°. Although the H0514-MP did produce some distortions around the edge of the image it was decided they were acceptable trade-off for the extra horizontal coverage that was obtained. Two Computar H0514-MP lenses have been used as the standard lenses for GIPSICAM surveys from the 2007/2008 season and onwards.
4.2.6.3 Video camera and camera lens set-up and positioning
The standard position for camera 1, referred to as the front camera, was the front centre mounting position. Here the camera is mounted in the centre of the vehicle, facing forward and parallel to the direction of travel, at a height of 2.3m above the ground, and facing down at the road at an angle of 4° from horizontal. Initially attached to the front camera was a Fujinon DF6HA-1B 6mm wide-angle lens. However, in 2007 the DF6HA-1B was replaced by the Computar H0514-MP 5mm wide-angle lens to get more horizontal coverage of the road carriageway.
The standard position for camera 2, referred to as the side camera, was the side mounting position. Here the camera is mounted looking out the side camera window, facing 60° to the left of the direction of travel, at a height of 2.3m above the ground, and facing down at the ground at an angle of 4° from horizontal. Initially attached to the side camera was a Fujinon DF6HA-1B 6mm wide-angle lens. However, in 2007 the DF6HA-1B was replaced by the Computar H0514-MP 5mm wide-angle lens to get more horizontal coverage of the side of the road.
Camera 3 is a special survey camera and is not used for standard surveys. The position of camera 3 is the front left mounting position. An example special survey is to video
107 Equipment Selection and Systems Integration
the road signs. Here the camera is mounted on the left side of the vehicle, facing 30° to the left of the direction of travel, at a height of 2.3m above the ground, and facing horizontal to the road. Attached to camera 3 is the Tamron 23FM25SP 25mm lens.
Camera 4 is a special survey camera and is not used for standard surveys. The position of camera 4 is the front right mounting position. An example special survey is to video the left side road shoulder. Here the camera is mounted on the right side of the vehicle, facing 45° to the left of the direction of travel, at a height of 2.3m above the ground, and facing down at the ground at an angle of 30° downwards from horizontal. Attached to camera 4 is the Fujinon HF9HA-1B 9mm lens.
Cameras 3 and 4 can be altered as required to suit the special survey. The design of the camera mounts enables the cameras’ configuration to be easily changed and is preset to allow easy reproducibility of settings for further surveys (See section 3.3.6 Camera turret).
Figure 61. GCv3 standard camera HFOV (horizontal plane relative to direction of travel)
108 Equipment Selection and Systems Integration
4.2.7 GCv3 onboard computers and data loggers
Within GCv2 there were a number of laptops and data loggers connected to equipment in order to collect data. This design was a result of the ad-hoc nature in which the system evolved over time. In the case of GCv3, the initial thought was to bring all of the processing power together on a single computer. However, one of the procedural inefficiencies with the operation of GCv2 was that the GPS data collection systems would be manually set to start collecting data at the start of a run and then manually closed (and data saved) at the end of a run, which could take 5 minutes to do. On open roads this could be an hour apart but in urban areas this could be as frequently as every 10 minutes. Thus each day of survey involved a significant amount of “overhead” just starting and stopping the capture of data from GPS receivers. It was decided to leave both the primary and secondary GPS receivers running continuously all day collecting data.
However, having the two GPS receivers running continuously meant that they should not be writing data to a single machine, also responsible for the other operations and video data recording. Also, given that the secondary GPS is a redundant system for the primary GPS, then it should have an independent data logger. The optimal solution therefore was to have a main computer to provide process/control for all survey activities, except for the primary and secondary GPS, which would have their own independent data logging device.
4.2.7.1 DELL Precision 670 dual processor server
The computer and parts procured, assembled and installed in the GCv3 vehicle was a DELL Precision 670 with the following specifications: 2 x Intel® Xeon™ 3.2GHz Processors 2GB ECC DDR2 SDRAM 256MB PCIe x16 NVIDIA Quadro FX 3400
109 Equipment Selection and Systems Integration
2 x 500GB SATA HDD 16x dual layer DVD+/-RW 2 x PCIe (x16/x8), 3 x PCIx and 1 x PCI slot QuikFire IDT804PCI (4 x FW800) 15” flatscreen, flexible keyboard and trackball
The operating system installed on the DELL Precision 670 was Microsoft Windows XP SP1.
The DELL Precision 670 is securely mounted in one of the vibration-proof shelves of the computer/equipment rack in the cargo area of the vehicle (See section 3.3.7 Computer/equipment rack). The computer is accessed via the operator console in the front of the vehicle (See section 3.3.8 Equipment operator work area).
4.2.7.2 DELL Latitude laptop as secondary GPS data logger
The secondary GPS data logger was a laptop. This had the advantage that the vehicle could be occasionally stopped for periods of time, such as at lunch time, when the GPS would still be logging data (and the data logger will still be expected to operate). This is where the advantage of the laptop comes in as a laptop runs off a battery. The vehicle can be turned off and the equipment powered down (except the two GPS, GIPSITRAC, and the FOG gyroscope), and the laptop could continue running on its internal battery.
A DELL Latitude laptop running Microsoft Windows 2000 was procured and securely mounted in the computer/equipment rack in the cargo area of the vehicle, for use as a data logger for the secondary GPS (See section 3.3.7 Computer/equipment rack).
4.2.7.3 Trimble Recon as primary GPS data logger
As with the secondary GPS data logger, a powerful computer was not required. In addition it was felt that the data logger for the primary GPS should be mounted in the
110 Equipment Selection and Systems Integration
front of the vehicle in order that the operator could monitor it. A Trimble Recon handheld data logger was used in GCv2 and it performed very well. In addition, the Trimble Recon runs off a battery, like a laptop, so could continue collecting data while the vehicle is turned off. A Trimble Recon handheld running Microsoft Windows Mobile was procured and securely mounted it in the wooden stationary compartment of the equipment operator work area (See section 3.3.8 Equipment operator work area).
4.2.8 Data storage
The GCv2 vehicle recorded video data directly onto mini DV tapes. However, recording directly onto a mini DV tape limits the length of a survey that can be undertaken. This in turn decreases the efficiency of the surveys as the vehicle is always stopping and performing end-of-run and start-of-run procedures instead of centreline surveying (See section 4.2.6 Video equipment).
A better solution was to record the video data onto a hard drive, which meant that the length of the survey would be limited only by the size of the hard drive space available. Testing revealed that the video data was streamed from the Sony DFW-SX910, and via the vehicle data capture software, at an average rate of approximately 1GB per camera per minute. With two 500GB HDD’s in the DELL Precision 670 computer more than 8 hours of survey time was possible before the data needed to be backed up. However, backing up the data from the internal HDD’s took hours, even using IEEE 1394b (FireWire 800) technology. The preferred solution therefore was to write the video data directly onto external HDD’s during surveys. Thus a HDD was to be used as a replaceable media device for the storage of the collected survey data.
In 2005 the four common personal computer high-speed serial connection technologies were USB, USB2, IEEE 1394a (FireWire 400) and IEEE 1394b (FireWire 800). IEEE 1394b offered the highest transfer speed of 800Mb/s. The only supplier who provided relatively low cost, large capacity external HDD’s with an IEEE 1394b interface was LaCie (pronounced “La See”). LaCie sold the 500GB LaCie Big Disk and the 1TB LaCie Bigger Disk, both available with a triple interface (USB2, 1394a and 1394b). The
111 Equipment Selection and Systems Integration
LaCie Big Disk had an external metal case containing two 250GB HDD’s which appeared to the computer as a single disk of capacity 500GB via the use of RAID 0 technology. The LaCie Bigger Disk looked like two Big Disks moulded together and contained four 250GB HDD’s which appeared to the computer as a single disk of capacity 1TB via the use of RAID 0 technology.
Twenty 500GB LaCie Big Disks and five 1TB LaCie Bigger Disks were procured. 5TB provided the GCv3 vehicle with one week of data storage. The external HDD’s were swapped back and forward between collecting data in the GCv3 vehicle and processing the data back in the office.
The 500GB drives were a good encapsulation of 4 hours survey work, representing half a day, and were preferred over the ‘chunkier’ but less reliable 1TB Bigger Disks. Over time LaCie introduced a number of new devices with capacities up to 4TB, which were also much more reliable.
4.3 Equipment integration
The data collection equipment within the GCv3 vehicle form subsystems that need to be synchronised and integrated to provide the necessary output data products.
4.3.1 Primary absolute position determination subsystem
4.3.1.1 Subsystem description
The subsystem comprises of a Trimble Pro XRS DGPS, Trimble Pro XRS antenna and a Trimble Recon handheld for data logging (See sections 4.2.1 DGPS receiver with real- time differential corrections and 4.2.7.3 Trimble Recon as primary GPS data logger). The Trimble Pro XRS antenna is mounted on the roof of the GCv3 vehicle, in the centre mount position (See section 3.3.10 GPS antenna mount points). The antenna cable feeds back into the GCv3 vehicle to the Trimble Pro XRS DGPS. The Trimble Pro XRS
112 Equipment Selection and Systems Integration
DGPS has OmniSTAR VBS DGPS service technology built-in and is thus able to receive and process the DGPS corrections transmitted by the OmniSTAR VBS service. Using the OmniSTAR VBS service the Trimble Pro XRS DGPS can provide real time positions accurate to within 1m (2D) (See section 2.3.1.3 OmniSTAR real-time DGPS). The Trimble Pro XRS DGPS is connected to a Recon handheld via a serial cable running the Trimble TerraSync software. The corrected DGPS data from the Trimble Pro XRS DGPS is stored on the Recon handheld.
Figure 62 Primary position determination system
4.3.1.2 Subsystem calibration
There is no calibration procedure for the Trimble Pro XRS DGPS. The position of the Trimble Pro XRS antenna is fixed within the GCv3 vehicle frame of reference and does not need to be recalibrated.
113 Equipment Selection and Systems Integration
4.3.1.3 Subsystem operational output
The Trimble Pro XRS DGPS receives both the GPS data and OmniSTAR VBS DGPS corrections, and calculates the real time DGPS data that is sent via serial cable to the Recon. The TerraSync software receives the input DGPS data and saves the data as Trimble observation files with the file extension .OBS.
4.3.2 Backup absolute position determination subsystem
The function of the backup absolute position determination subsystem is to provide a fallback system that can be utilised in situations when the primary position determination subsystem has failed, or may be producing erroneous data.
4.3.2.1 Subsystem description
The subsystem comprises of two parts, the GPS base station at the RTA office at Rockdale and a GPS rover within the GCv3 vehicle.
4.3.2.1.1 GPS base station
The GPS base station system comprises a Trimble Pro XL GPS receiver, Trimble PFCBS ground-plane antenna and a PC running Trimble’s Pathfinder Community Base Station (PFCBS) v2.68 software. The PFCBS ground-plane antenna is placed on the roof of the building at the Rockdale Works Office. The antenna cable feeds back inside the building to the Trimble Pro XL GPS receiver. The Trimble Pro XL GPS receiver is connected to a PC via a serial cable. The PC runs the PFCBS software collecting synchronised measurements that are saved as SSF format files. The Trimble SSF (Standard Storage Format) file format is a proprietary binary format, however Trimble does provide tools to export the data to other formats.
114 Equipment Selection and Systems Integration
Figure 63 Community Base Station hardware setup (Trimble, 1997)
4.3.2.1.2 GPS rover
The GPS rover system comprises a Trimble Pro XL GPS receiver, Trimble Pro XL antenna and a laptop running Trimble’s ASPEN v2.11 software. The Trimble Pro XL antenna is mounted on the roof of the GCv3 vehicle, in the right hand side (driver side) mount position (See section 3.3.10 GPS antenna mount points). The antenna cable feeds back into the GCv3 vehicle to the Trimble Pro XL GPS. The Trimble Pro XL GPS is connected to an Intel-based laptop via a serial cable. The laptop is running ASPEN software collecting synchronised measurements that are saved as SSF format files.
4.3.2.1.3 Post-processing the GPS rover data with the GPS base station data
The synchronised measurements collected from the GPS rover and the GPS base station are post-processed in the office using Trimble’s GPS Pathfinder Office v3.0 software and double-differencing software developed by the RTA. Post-processing the GPS rover data can provide positions accurate to within 1m (2D).
115 Equipment Selection and Systems Integration
4.3.2.2 Subsystem calibration
There is no calibration procedure for either GPS receiver. The position of the Trimble Pro XL rover antenna is fixed within the GCv3 vehicle frame of reference and is not recalibrated. The position of the PFCBS ground-plane antenna was surveyed when initially set-up.
4.3.2.3 Subsystem operational output
The Trimble Pro XL base station GPS data are saved as .SSF files by the PFCBS software. The Trimble Pro XL rover GPS data are saved as .SSF files by the ASPEN software.
4.3.3 GCAM subsystem
The function of the GCAM subsystem is to provide: Continuous, accurate relative position of the GIPSICAM vehicle in real time Continuous video from up to four video cameras Continuous, sensor data for calculation of road geometry information
The real time relative position data are used to supplement the position data in areas of no GPS signal such as in tunnels, or in areas causing problems for GPS signals such as the Sydney CBD or in mountainous terrain where multipath and canyoning will be present respectively. The video is combined with the vehicle position data to produce .JPG images at known positions and defined distance intervals along the road pavement. The sensor data is combined with vehicle position data to calculate road geometry information, consisting of grade, crossfall, horizontal radius and vertical radius at known positions and defined distance intervals along the road pavement.
116 Equipment Selection and Systems Integration
4.3.3.1 Subsystem description
The subsystem comprises of three parts, performing the three functions of collecting relative position data, video and sensor data. The primary components consist of a DELL Precision 670 computer, an ARRB GIPSITRAC, a KVH DSP-3000 FOG and the Sony DFW-SX910 video cameras. The GCAM software running on the computer creates an ASCII synchronisation file called .GPSTIMES which contains a GPS time- stamp expressed in GPS seconds, a computer time-stamp expressed in seconds, and a GIPSITRAC TT record count that indicates distance travelled. Sample .GPSTIMES file (truncated to first 10 rows):
PC Times GPS Seconds Tt Count 326.337003 188231.000 2 327.337197 188232.000 3 328.336910 188233.000 4 329.336924 188234.000 5 330.336825 188235.000 6 331.336819 188236.000 7 332.336744 188237.000 8 333.336706 188238.000 9 334.336654 188239.000 10 335.336620 188240.000 11
4.3.3.1.1 Relative positioning data
The relative positioning data collection components consist of a KVH DSP-3000 FOG, an ARRB GIPSITRAC, and a DELL Precision 670 computer. The DSP-3000 is mounted in the equipment/storage cabinet in the rear cargo area of the vehicle (See section 3.3.1 Equipment/storage cabinet). The DSP-3000 is connected to the computer via a serial cable. The GCAM software on the computer time-stamps the incoming DSP-3000 data and saves it in an ASCII .DSP3000 file. The .DSP3000 file contains the gyroscope turn-rate values and a computer time-stamp (expressed in seconds). Sample .DSP3000 file (truncated to first 10 rows):
326.095 -0.009941 326.097 -0.002812 326.100 -0.016910 326.101 -0.000252
117 Equipment Selection and Systems Integration
326.102 -0.010251 326.104 -0.000795 326.105 -0.003508 326.107 -0.006902 326.108 -0.006475 326.138 0.000921
The GIPSITRAC is connected to the computer via a serial cable and sends GIPSITRAC messages that transfer inertial data collected from the GIPSITRAC sensors (See section 4.3.3.1.3 GIPSITRAC sensor data).
4.3.3.1.2 Video data
The video data collection components consist of the Sony DFW-SX910 video cameras and the DELL Precision 670 computer. The DFW-SX910 video cameras are connected to the computer via IEEE 1394a to IEEE 1394b cables. The GCAM software on the computer time-stamps the frames captured from the video cameras and stores the synchronisation data in ASCII .TIM files. The video is stored in a binary .AVI file. A matching .AVI file and .TIM exists for each camera used during the survey. The .TIM file contains the camera frame number and the computer time-stamp (expressed in seconds). Sample .TIM file (truncated to the first 10 rows):
0.0 326.080 1.0 326.114 2.0 326.246 3.0 326.379 4.0 326.514 5.0 326.647 6.0 326.780 7.0 326.913 8.0 327.046 9.0 327.180
4.3.3.1.3 GIPSITRAC sensor data
The GIPSITRAC sensor data collection components consist of an ARRB GIPSITRAC, a Motorola patch antenna, a Philips Semiconductors KMI 15/1 rotational speed sensor,
118 Equipment Selection and Systems Integration
a DELL Precision 670 computer running Windows XP, and the RTA-developed GCAM software.
The Motorola patch antenna is mounted on the roof of the GCv3 vehicle, in the left hand side (Passenger side) mount position (See section 3.3.10 GPS antenna mount points). The antenna cable feeds back into the GCv3 vehicle to the Motorola PVT6 GPS inside the GIPSITRAC.
The Philips Semiconductors KMI 15/1 sensor unit is attached to the GCv3 vehicle differential housing and interacts with a steel cog attached to the GCv3 vehicle drive shaft (See section 3.3.9 GIPSITRAC odometer). The steel cog has 40 teeth giving a resolution of 40 pulses per drive shaft revolution. The sensor unit cable feeds back into the GCv3 vehicle to the GIPSITRAC.
The GIPSITRAC consists of two Honeywell QA700 accelerometers, two Murata ENV05S gyroscopes, a Motorola PVT6 GPS and a microprocessor (ARRB, 1995). The Honeywell QA700 accelerometers are placed in a longitudinal and a transverse orientation with respect to the front of the GCv3 vehicle. The longitudinal accelerometer measures road grade. The transverse accelerometer measures road crossfall. Horizontal and vertical radius can also be calculated from the accelerometers. The Murata ENV05S gyroscopes are placed in an upright and an up-down orientation. The gyroscopes determine the direction of travel. There are two gyroscopes as a form of redundancy, hence it is possible to average the data from both.
The GIPSITRAC is connected via a serial cable to the DELL Precision 670 computer, running the RTA’s GCAM software, which collects the sensor data output by the GIPSITRAC. The GIPSITRAC receives input pulses from the PVT6 GPS and the KMI 15/1 odometer. The GIPSITRAC uses these pulses to integrate and control the sampling of the GIPSITRAC sensors. GIPSITRAC then sends formatted ARRB GIPSITRAC messages (records) to the computer running the GCAM software. The GCAM software then writes these messages to two binary files. The first binary file contains only the PVT6 GPS data and has a file extension of .GPS. The second binary file contains all other GIPSITRAC sensor data plus the first two minutes of the GPS data, and has a file
119 Equipment Selection and Systems Integration
extension of .BIN. At the end of a survey run the GCAM software copies the .BIN file to an .INE file and appends an ASCII header to the front of the .INE file containing information needed to process the .INE file (such as configuration details and survey details).
An input ASCII Control Point File, with the file extension .CP, is utilised by the GCAM software to determine control points along a survey run. The control points specify the start and end points of the road sections, as defined by the RTA RoadLoc linear reference system (See section 2.4.3 RTA RoadLoc). The .CP files are generated using RTA software in the office before going out into the field to commence a survey. A separate .CP file is created for each state road.
4.3.3.1.3.1 GIPSITRAC BG messages
The PVT6 GPS generates GPS data every one second and sends the data as a pulse to the GIPSITRAC microprocessor which then sends a GIPSITRAC BG message via serial cable to the computer running the GCAM software. The GCAM software saves all BG messages into the .GPS file. The first 30 seconds of BG messages are also saved into the .BIN file. The BG records contain the satellite psuedorange data at a 1-second interval epoch. Sample GIPSITRAC BG record (converted to ASCII):
Bg 188329.000516111 13 08 188328.932888202 20147.7787323 176053 1612 23 08 188328.927768574 23143.0316010 364588 -1098 27 08 188328.931905278 29343.1892090 149159 1446 3 88 188328.923035609 22301.1914215 1519228 2420 8 08 188328.925714074 54973.8000183 450448 -1220 19 88 188328.924604139 36875.6494904 623659 -1499
4.3.3.1.3.2 GIPSITRAC BA messages
A PVT6 GPS pulse to the GIPSITRAC microprocessor also triggers the sending of a GIPSITRAC BA message, which the GCAM software on the computer saves in the .BIN file. The BA records contain the predicted position of the GIPSITRAC system at “one second into the future” based on the current satellite psuedorange data and taking
120 Equipment Selection and Systems Integration
into account the current bearing and distance (speed) information from the ENV05S gyroscopes and the KMI 15/1 odometer. Sample GIPSITRAC BA record (converted to ASCII):
Ba 2006/02/07 14:18:50.000529133 -33.95143778 151.16013250 18.18 -1.37 2.8 162.2 2.3 P 7 7 13 08 67 A0 23 08 170 A0 27 08 177 A0 3 08 72 A0 8 08 145 A0 19 08 42 A1 20
4.3.3.1.3.3 GIPSITRAC TT messages
A PVT6 GPS pulse to the GIPSITRAC microprocessor also triggers the sending of a GIPSITRAC TT message, which the GCAM software on the computer saves in the .BIN file. The TT records contain timing information based on distance pulses received from the KMI 15/1 odometer. Each TT record contains the total coarse-odometer count and the current intermediate fine-odometer count for each GPS pulse. Sample GIPSITRAC TT record (converted to ASCII):
Tt 26 52 101
4.3.3.1.3.4 GIPSITRAC RR messages
The KMI 15/1 odometer sends a pulse as each tooth on the steel cog passes by the sensor. The steel cog installed in-line with the GCv3 vehicle drive shaft has 40 teeth, thus 40 pulses are sent for one rotation of the drive shaft. Each of these 40 pulses is referred to as a fine-odometer pulse. However, travelling at high (legal) speeds such as 80km/h means that the rate of fine-odometer pulses sent is very large, which will overwhelm the 9600 bps serial connection to the computer when trying to send GIPSITRAC messages for every fine-odometer pulse. The solution is the coarse- odometer count, which is a defined unit of distance that is used to synchronise and integrate the GIPSITRAC sensors. A coarse-odometer distance of approximately 2m is
121 Equipment Selection and Systems Integration
recommended by ARRB. Experimentation with the 40 tooth cog on the GCv3 vehicle revealed that a total of around 150 teeth corresponded to a distance of approximately 2m, depending on the size of the tyres, the wear of the tyres, how much air was in the tyres and if the tyres were warmed up. The GIPSITRAC odometer calibration is undertaken regularly to ensure the coarse-odometer distance is known (See section 6.2.5 GIPSITRAC odometer calibration). The coarse-odometer count and the coarse- odometer distance are recorded in the header of the .INE file. For example, after a GIPSITRAC odometer calibration undertaken in January 2006, and using a fine- odometer count of 150, the calibrated coarse-odometer distance was 1.9231 metres. Every 150 fine-odometer pulses corresponds to a coarse-odometer pulse that triggers the sending of a GIPSITRAC RR message, which the GCAM software on the computer saves in the .BIN file. The RR records contain the measured data from the QA700 accelerometers and ENV05S gyroscopes. Each RR record contains the total coarse- odometer count, the transverse and longitudinal accelerometer readings, the two gyro readings and timing information between the current and the last RR record. Sample GIPSITRAC RR record (converted to ASCII):
Rr 26 -7624 -13373 44045 -62375 58 578926
4.3.3.1.3.5 GIPSITRAC SS messages
If the GIPSITRAC stops receiving the KMI 15/1 odometer pulses then it assumes that the vehicle has stopped, which triggers the sending of a GIPSITRAC SS message, which the GCAM software on the computer saves in the .BIN file. The SS records contain the measured data from the QA700 accelerometers and ENV05S gyroscopes. Each SS record contains the total coarse-odometer count, the current fine-odometer count, the transverse and longitudinal accelerometer readings, the two gyro readings and timing information between the current SS record and the last RR record. Sample GIPSITRAC SS record (converted to ASCII):
Ss 668 65 24239 2995 -4267 -69938 191
122 Equipment Selection and Systems Integration
4.3.3.1.3.6 GIPSITRAC MM messages
If, after being stationary, the GIPSITRAC starts receiving the KMI 15/1 odometer pulses again then it assumes that the vehicle has started moving which triggers the sending of a GIPSITRAC MM message, which the GCAM software on the computer saves in the .BIN file. The MM records contain the measured data from the QA700 accelerometers and ENV05S gyroscopes. Each MM record contains the total coarse- odometer count, the current fine-odometer count, the transverse and longitudinal accelerometer readings, the two gyro readings and timing information between the current MM record and the last RR record. Sample GIPSITRAC MM record (converted to ASCII):
Mm 750 9 194026 279207 -94788 -1577988 4313
4.3.3.1.3.7 GIPSITRAC II messages
If the GIPSITRAC receives the KMI 15/1 odometer pulses at a very slow rate indicating that the vehicle is moving extremely slowly, but is not stationary, then this triggers the sending of a GIPSITRAC II message, which the GCAM software on the computer saves in the .BIN file. The II records contain the measured data from the QA700 accelerometers and ENV05S gyroscopes. Each II record contains the total coarse- odometer count, the current fine-odometer count, the transverse and longitudinal accelerometer readings, the two gyro readings and timing information between the current II record and the last RR record. Sample GIPSITRAC II record (converted to ASCII):
Ii 1665 22 240247 98638 -96119 -1646893 4499
4.3.3.1.3.8 GIPSITRAC EE messages
The GIPSITRAC indicates an error by the sending of a GIPSITRAC EE message, which the GCAM software on the computer saves in the .BIN file. The EE records
123 Equipment Selection and Systems Integration
contain a code indicating the error that has occurred. Sample ARRB EE record (converted to ASCII):
Ee Error = 3
4.3.3.1.3.9 GIPSITRAC CC messages
The GCAM software operator may save comments during a survey in the form of a GIPSITRAC CC message, which the GCAM software on the computer saves in the .BIN file. In general the CC record is used to indicate the location of features along a road. Sample GIPSITRAC CC records (converted to ASCII):
Cc 0 149 Start Run Cc 675 37 |0010|0.000|A1|0.000|The Grand Parade (MR 194):|1386825|20894 Cc 1170 19 |0020|0.949|A1|0.949|West Botany St (SR 2032): |1386835|20895 Cc 1693 0 |0030|0.997|A1|1.946|Princes Hwy (SH 1): |1386848|77693 Cc 2878 11 Missed CP
4.3.3.1.3.10 GCAM .INE file
At the end of every survey run the GCAM software copies the .BIN file to an .INE file and then appends an ASCII header record to the front of the .INE file. The header record contains information concerning the section of road being surveyed, the date and time the survey was conducted, and configuration and calibration information used for the survey run. Sample .INE file header record:
VERSION | 2 GPS START WEEK | 1403 GPS START SECONDS | 124833 PC START DATE | 27/11/06 PC START TIME | 10:40:20 RUN NAME | 00169AND START LINK | 0010 LAST LINK | 0130 OPERATOR | Dennis Entriken DRIVER | Mike Vernon MASK | NO GPS END WEEK | 1403 GPS END SECONDS | 126048 PC END DATE | 27/11/06
124 Equipment Selection and Systems Integration
PC END TIME | 11:00:37 COMMENTS | Overcast sky with intermittent light rain GPS1 DETAILS | Name=Primary, X=1.500 , Y=0.000 , Z=2.480 GPS2 DETAILS | Name=Secondary, X=1.240 , Y=0.360 , Z=2.500 GPS3 DETAILS | Name=Gipsitrac, X=1.240 , Y=-0.360, Z=2.510 CAMERA A DETAILS | CN=1,X=2.850,Y=0.000,Z=2.270,FrameRate=7.5,CaptureInterval=1 CAMERA B DETAILS | CN=4,X=0.000,Y=0.000,Z=0.000,FrameRate=7.5,CaptureInterval=1 CAMERA C DETAILS | CN=3,X=0.000,Y=0.000,Z=0.000,FrameRate=7.5,CaptureInterval=1 CAMERA D DETAILS | CN=2,X=0.000,Y=0.000,Z=0.000,FrameRate=7.5,CaptureInterval=1 CALIBRATION NAME | GCv3_Nov_2006 CAMERA A CALIB NAME | a CAMERA B CALIB NAME | b CAMERA C CALIB NAME | c CAMERA D CALIB NAME | d COURSE ODO DISTANCE | 1.9231 FINE ODO COUNT | 150 GYRO 1 ZERO OFFSET | -11.8652 GYRO 1 SENSITIVITY | -2351.3442 GYRO 2 ZERO OFFSET | -350.7210 GYRO 2 SENSITIVITY | 2124.6000 TRANSVERSE ZERO OFFSET | -10.1633 TRANSVERSE SENSITIVITY | -187.4944 LONGITUDINAL ZERO OFFSET | 37.0657 LONGITUDINAL SENSITIVITY | -108.5364 TIME ZONE HOURS | 10.00 PITCH FACTOR | 0.9400 ROLL FACTOR | 0.8600 LONG SPRING CALIBRATION | -18.93 -19.37 LONG SPRING CALIBRATION | -10.30 -10.55 LONG SPRING CALIBRATION | -6.68 -6.94 LONG SPRING CALIBRATION | 6.68 7.21 LONG SPRING CALIBRATION | 10.30 10.75 LONG SPRING CALIBRATION | 18.93 19.98 TRAN SPRING CALIBRATION | -30.00 -32.81 TRAN SPRING CALIBRATION | -25.00 -27.26 TRAN SPRING CALIBRATION | -20.00 -21.79 TRAN SPRING CALIBRATION | -15.00 -16.33 TRAN SPRING CALIBRATION | -10.00 -10.88 TRAN SPRING CALIBRATION | -5.00 -5.40 TRAN SPRING CALIBRATION | -4.00 -4.29 TRAN SPRING CALIBRATION | -3.10 -3.33 TRAN SPRING CALIBRATION | -1.80 -1.81 TRAN SPRING CALIBRATION | 2.00 1.87 TRAN SPRING CALIBRATION | 3.00 3.26 TRAN SPRING CALIBRATION | 4.00 4.35 TRAN SPRING CALIBRATION | 5.00 5.45 TRAN SPRING CALIBRATION | 10.00 10.84 TRAN SPRING CALIBRATION | 15.00 16.29 TRAN SPRING CALIBRATION | 20.00 21.76 TRAN SPRING CALIBRATION | 25.00 27.29 TRAN SPRING CALIBRATION | 30.00 32.61 KVH DSP 3000 USED | YES END HEADER
4.3.3.1.3.11 Synchronisation and integration of GIPSITRAC messages
125 Equipment Selection and Systems Integration
The sending of the TT, BA and BG messages is triggered by the PVT6 GPS pulse, which is generated every second. The order of the generated messages is TT, BA and BG. Thus every second a TT, BA and BG message will be sent by the GIPSITRAC in the above order. Sample GIPSITRAC TT, BA and BG records in sequence (converted to ASCII):
Tt 66 47 121 Ba 2006/02/07 14:19:10.000789537 -33.95159694 151.16093528 20.57 1.01 0.0 0.0 2.3 P 7 7 13 08 78 A0 23 08 200 A0 27 08 184 A0 3 08 141 A0 8 08 150 A0 19 08 142 A0 20 Bg 188349.000776515 13 08 188348.933146119 33253.4187012 1345152 2082 23 08 188348.928000148 60282.7167511 1575189 -898 27 08 188348.932178322 744.6303406 1294426 2035 3 08 188348.923289460 28999.1609497 1119314 2109 8 08 188348.925995739 39955.2444000 6715 -666 19 08 188348.924889131 27097.8000793 174685 -1853
The sending of the RR, SS, MM and II messages is triggered by the KMI 15/1 odometer pulse. The RR message is sent approximately every two metres, relating to the coarse- odometer distance, while the vehicle is in motion above a preset speed. If the vehicle stops then an SS message is sent. When the vehicle starts moving again an MM message is sent. If the vehicle is in motion below a preset speed then an intermediate II message is sent. A CC message can be generated by the operator, GCAM or GIPSITRAC at any time. An EE message is generated whenever an error occurs.
4.3.3.1.3.12 Control Point file
An ASCII Control Point file is used by the operator to indicate control points along a survey run. This is achieved within the GCAM software via the use of a GIPSITRAC CC record, which is saved in the binary .BIN file, and copied to the .INE file at the end of the survey run. The use of the CC record in this way means that the operator can
126 Equipment Selection and Systems Integration
specify a control point such as the start of a new road section or a road feature, and this information is then integrated into the capture process via the coarse-odometer count and the fine-odometer count which translates to a chainage along the road and a spatial position after post-processing the data. The Control Point file for a road is created by RTA software, which extracts the control point data from the RTA RoadLoc database and saves this data as an ASCII file with the file extension of .CP. Sample .CP file for MR667 in the prescribed direction:
------LINK| LINK |CW| RUN | CP DESCRIPTION | RF_ID | LCID | VITC |FWD| -NO. | ODO |VS| DIST | & COMMENTS | | | |OFF| ------Section 0010 to 0020 ***** President Ave, Brighton-Le-Sands |0010| 0.000|A1| 0.000|The Grand Parade (Mr 194):|1418275| 76781|00:06:50.58|6.5 |0020| 1.061|A1| 1.061|West Botany St (Sr 2032): |1418284| 76782|00:08:23.24|6.5 |XXXX| 0.519| | 1.580|Princes Highway (Sh 1): |1418291| |00:11:38.84|6.5 END OF FILE
4.3.3.2 Subsystem calibration
The KVH DSP-3000 FOG has no user calibration. The front Sony DFW-SX910 video camera is calibrated according to the GCv3 camera/lens calibration procedure (See section 6.2.1 Camera/Lens calibration). There are a number of calibration procedures that are undertaken to ensure accurate data is produced from the GIPSITRAC sensors. The purpose of the calibration procedures is to determine the following unknowns (ARRB, 1995): Longitudinal accelerometer zero offset and sensitivity Transverse accelerometer zero offset and sensitivity Gyroscope 1 yaw rate zero offset and sensitivity Gyroscope 2 yaw rate zero offset and sensitivity Odometer pulses per km
Refer to (section 6.2 Calibration procedures) for more details.
127 Equipment Selection and Systems Integration
4.3.3.3 Subsystem operational output
The KVH DSP-3000 FOG data and synchronisation data are saved as .DSP3000 files. The Sony DFW-SX910 video data are saved as .AVI files. Matching .TIM files providing synchronisation data are also created. The ARRB GIPSITRAC data are saved as .GPS, .BIN, and .INE files. The GCAM software saves GPS, computer, and odometer synchronisation data in a .GPSTIMES file.
4.3.4 Integration of the subsystems
The .GPSTIMES file enables the synchronisation of the data from each of the subsystems. The computer timestamp field in the .GPSTIMES file is synchronised with the computer timestamp field in the .DSP3000 file and the computer timestamp field in the .TIM file which enables the integration of the KVH DSP-3000 FOG gyroscope turn- rate values and the Sony DFW-SX910 video frames. The GIPSITRAC TT record count field in the .GPSTIMES file is synchronised with the GIPSITRAC TT record count in the .INE file which enables the integration of the ARRB GIPSITRAC sensor data. The GPS seconds field in the .GPSTIMES file is synchronised with the GPS seconds data in .OBS file and the GPS seconds data in the .SSF file which enables the integration of the Trimble Pro XRS DGPS data and the Trimble Pro XL GPS data. Thus, all of the sensor data is integrated by synchronising with the .GPSTIMES file.
4.4 Concluding remarks
The GCv3 vehicle has now been operational for five seasons. Overall, the selection and integration of the equipment and sensors during the development of GCv3 was highly successful and aligned very well with the three primary design considerations for GCv3, which were: quality/accuracy, efficiency and OH&S.
However, with the advancement of technology each year the MMS equipment and sensors get better and cheaper. As with all assets, the GIPSICAM technology needs to
128 Equipment Selection and Systems Integration
be not just maintained but routinely upgraded to keep up with the current and future information requirements of the RTA. Incorporating newer and additional equipment and sensors will ensure the GIPSICAM technology continues to be an important technology providing vital information within the RTA.
The primary GPS initially utilised in GCv3 was the Trimble Pro XRS DGPS. A firmware update ensured the equipment was performing optimally, and its performance exceeded our primary GPS positioning requirements. In mid-2007 a Trimble Pro XRT GPS was procured to replace the Trimble Pro XRS DGPS, but this was primarily the result of a funding opportunity and not a performance deficiency of the Trimble Pro XRS DGPS. The Trimble Pro XRT GPS did however provide the capability to experiment with the OmniSTAR HP service, though unfortunately the OmniSTAR HP service was found to be unreliable under tree canopies which meant it was unsuitable for many roads that traversed national parks and state forests.
With the current and future availability of other GNSS such as GLONASS, Galileo and COMPASS, the situation is coming where up to 100 GNSS satellites could be in orbit providing GNSS signals, with possibly up to 50 satellites visible at an epoch. Thus upgrading the primary and secondary GPS receivers to receivers that can receive signals from multiple GNSS to determine a position should increase the reliability and accuracy of the GIPSICAM absolute position capability.
The ARRB GIPSITRAC provides a good INS and road geometry capability; however its technology is limited to two accelerometers, two gyroscopes and a six channel GPS receiver. Replacing GIPSITRAC with current GPS/INS technology that utilises three accelerometers and three gyroscopes, in a 3-axis configuration, with a multi-GNSS capable DGPS, should increase the reliability and accuracy of the GIPSICAM relative positioning capability and the road geometry sensing capability. Alternatively, upgrading the components and configuration of GIPSITRAC to similar specifications as above could also increase the reliability and accuracy of the GIPSICAM relative positioning capability and the road geometry sensing capability.
129 Equipment Selection and Systems Integration
The daily GIPSITRAC calibration procedures are an area of inefficiency. The GCv3 vehicle staff undertake the daily GIPSITRAC calibration procedures each morning, which takes considerable time to first find a suitable area to undertake the calibration, and then to actually do the calibrations. Daily GIPSITRAC calibration procedures can take between 15 to 30 minutes to perform, often longer for new and inexperienced GCv3 vehicle staff. Removing the need for the daily calibrations would increase efficiency of the surveys. Investigation into the replacement of GIPSITRAC components or GIPSITRAC itself with technology that does not require calibration, such as fibre optic gyroscopes, should lead to increases in efficiency.
The Sony DFW-SX910 cameras perform well; however higher resolution digital video cameras with faster frame rates are now available. Additionally, advancements in digital video camera CCD technologies are solving the visual artefacts issues found in the Sony DFW-SX910 camera such as vertical smearing and blooming (See section 7.5.2 Hardware related road image quality problems). Research into utilisation of a calibrated panoramic camera mounted above the camera turret could possibly lead to new capabilities for GIPSICAM. A review of current digital video camera technology and an upgrade of the GCv3 video technology should lead to increases in quality, accuracy and capability for GIPSICAM.
Research into sensors to collect new data types could also increase the capability of the GIPSICAM technology. The utilisation of lasers to profile the road pavement and determine the horizontal/vertical clearances around the road corridor would prove to be useful. The addition of a Ball-Bank meter to determine and record state-wide advisory speeds data would be a useful capability. LiDAR is another technology that should be researched and incorporated into the GIPSICAM technology. LiDAR technology used in conjunction with video/GPS/INS technology has the potential to provide a more accurate and efficient automated asset inventory capability that far surpasses using video/GPS/INS technology alone.
For other suggested upgrades to the current GIPSICAM technology see section 9.2 Future research and development opportunities for GIPSICAM.
130 Software
CHAPTER 5 SOFTWARE
5.1 Introduction
The NSW Roads and Traffic Authority (RTA) has developed a suite of GIPSICAM software to enable the capture, processing and display of the GIPSICAM road image, centreline and road geometry data. In addition, the RTA also utilises commercial software such as the ESRI ArcGIS suite to display spatial data.
5.1 GCv3 vehicle software
5.1.1 GIPSICAM GCAM
GCAM is short for ‘GIPSICAM’, and is name given to the in-vehicle GCv3 software (See Figure 64, Figure 65 and Figure 66). GCAM was developed as part of the GCv3 project and replaces the old DOS-based ARRB GIPSITRK software that was used by GCv2 and GCv1. GCAM performs the following in-vehicle functions: GCv3 survey control, sensor synchronisation and data acquisition. GIPSITRAC bench calibration (6.2.2 GIPSITRAC bench calibration) GIPSITRAC odometer calibration (6.2.5 GIPSITRAC odometer calibration) GIPSITRAC daily 180 calibration (6.2.6 GIPSITRAC daily 180 calibration) GIPSITRAC daily 360 calibration (6.2.7 GIPSITRAC daily 360 calibration) Camera/lens calibration grid capture (6.2.1 Camera/Lens calibration) GCv3 suspension calibration (6.2.4 GIPSITRAC suspension/tilt calibration)
Figure 64. GCAM main menu
131 Software
Figure 65. GCAM Acquire module showing a survey in progress
Figure 66. GCAM Calibration module
132 Software
5.2 GCv3 processing software
5.2.1 GIPSICAM GCAPTURE
GCAPTURE is short for ‘GIPSICAM CAPTURE’, and is the GIPSICAM processing software that integrates the sensor data and produces the output GIPSICAM datasets (See Figure 67, Figure 68 and Figure 69). GCAPTURE was existing software that was enhanced as part of the GCv3 project.
Figure 67. GCAPTURE showing unadjusted DGPS/INS data (RTA, 2011 GCAP)
133 Software
Figure 68. GCAPTURE vertical adjustment (RTA, 2011 GCAP)
Figure 69. GCAPTURE video processing (RTA, 2011 GCAP)
134 Software
5.3 GCv3 asset capture software
5.3.1 GIPSICAM AssetLoc
AssetLoc is the GIPSICAM road image display and asset capture application. It can display road imagery from any of the four GCv3 cameras, measure width/length/height/area from road imagery, and manually capture asset positions and attributes from the road imagery (See Figure 70 and Figure 71). AssetLoc was existing software that was enhanced as part of the GCv3 project.
Figure 70. AssetLoc being used to measure the width of the lane at the Rockdale test area
Figure 71. GIPSICAM road image annotation
135 Software
5.4 GCv3 dataset display software
5.4.1 GIPSICAM RoadFlix
RoadFlix is the GIPSICAM road image data viewing application. It can display road imagery from any of the four GCv3 cameras, and has the ability to “goto” a RoadLoc reference (See Figure 72). RoadFlix was existing software that was enhanced to be able to display the multiple-view camera images.
Figure 72. RoadFlix showing the GIPSICAM road images
136 Software
5.4.2 GIPSICAM Road Geometry Analyst
Road Geometry Analyst (RGA) provides the capability to visualise and analyse the GIPSICAM road geometry data (See Figure 73). RGA did not require enhancement to utilise the GCv3 data.
Figure 73. RGA showing GIPSICAM road geometry data
137 Software
5.4.3 GIPSICAM RoadBrowser
RoadBrowser provides the capability to integrate the GIPSICAM road image data with orthophotography, topographic maps, ‘UBD’ and ‘Sydway’ street directories, and NSW state road centreline data. RoadBrowser also provides a search capability to locate by RoadLoc reference, coordinate, street address, intersection, and point of interest (See Figure 74). RoadBrowser did not require enhancement to utilise the GCv3 data.
Figure 74. RoadBrowser showing the location of test road MR667
138 Software
5.4.4 ESRI ArcGIS
The ESRI ArcGIS suite is the RTA’s corporate GIS. A GIPSICAM point feature class allows ArcGIS users to spatially display the positions of the GIPSICAM images and road geometry data, and the hyperlink functionality enables the display of the road images (See Figure 75). ArcGIS did not require enhancement to utilise the GCv3 data.
Figure 75. GIPSICAM image points displayed in ESRI ArcGIS on SKM orthophoto
139 Operational Procedures
CHAPTER 6 OPERATIONAL PROCEDURES
6.1 Introduction
As with a software application or a hardware device, a set of operational procedures is required for the effective operation of a complex system. In the case of data collection equipment such as GCv3 it is critical to be able to conduct multiple surveys of the same section of road and produce the same results. This consistency of results requires a high level of confidence in the output GIPSICAM data and in the operation of the overall GIPSICAM system.
In order to ensure consistency of results a set of operational procedures were developed. These procedures form part of the GIPSICAM Quality Management System (QMS), and cover calibration of the equipment, office and field routines, and validation of the output GIPSICAM data.
6.2 Calibration procedures
Calibration of the equipment in GCv3 is necessary to ensure consistent results. While some of the equipment, such as the Fibre Optic Gyro (FOG) and Differential Global Positioning System (DGPS) do not require user calibration, the GIPSITRAC instruments do require calibration because of the potential for inaccurate data or results if not carefully calibrated.
6.2.1 Camera/Lens calibration
The ability to measure and geocode from a GIPSICAM image is one of the important capabilities of GIPSICAM. Such a capability requires a means of converting from user
140 Operational Procedures
‘mouse clicks’ on a GIPSICAM image to ‘real world’ coordinates. To be confident of results camera and lens calibration is necessary.
6.2.1.1 Purpose
The purpose of the camera/lens calibration is to determine the conversion matrix between GIPSICAM image pixels and the vehicle reference frame. Vehicle reference system coordinates can then be transformed into real world coordinates by relating the vehicle reference frame to a geodetic reference frame. The camera/lens calibration is routinely undertaken annually at the start of each survey season. Ad hoc camera/lens calibration is undertaken whenever the front camera is removed/installed in the front camera mounting position.
6.2.1.2 Procedure summary
The camera lens to be utilised for the survey season is attached to the front camera body. Once installed and calibrated, the front camera cannot be dismantled or removed as doing so invalidates the calibration and a new calibration procedure must be undertaken. The GCv3 vehicle is positioned on level ground and perpendicular to the lens calibration wall. The lens calibration wall consists of a large grid and a number of marked points on a fixed wall. The height above the ground and the distance from the centre of the wall to each grid line and each marked point has been accurately measured.
First the GCv3 vehicle is positioned close to the lens calibration wall. The vehicle driver and operator then sit in their seats to account for the effect of their weight on the vehicle suspension, while an image of the lens calibration wall is captured by the front camera. The distance from the front camera to the lens calibration wall is accurately measured. The vehicle is then moved away from the lens calibration wall, and an image of the lens calibration wall is again captured and the distance from the front camera to the lens calibration wall is again accurately measured. All calibration details are recorded on a lens calibration field sheet.
141 Operational Procedures
Finally the “near” and “far” images of the lens calibration wall and the data from the lens calibration field sheet is loaded into software which outputs the conversion matrix. This conversion matrix information is then utilised during the processing of GIPSICAM surveys to enable the internal image georeferencing capability of the GIPSICAM system, which is fundamental to the asset capture functionality.
6.2.2 GIPSITRAC bench calibration
The GIPSITRAC box contains MEMS sensors that need to be calibrated so that the sensor outputs can be correctly and consistently interpreted. The GIPSITRAC bench calibration is one of a number of calibration procedures that are used to calibrate the MEMS sensors in the GIPSITRAC box.
6.2.2.1 Purpose
The purpose of the GIPSITRAC bench calibration is to determine the sensitivity constants for the longitudinal and transverse accelerometers within the GIPSITRAC box (ARRB, 1995), to enable the interpretation of the accelerometer outputs. The GIPSITRAC bench calibration is routinely carried out annually at the start of each survey season. Ad hoc GIPSITRAC bench calibration is undertaken whenever the GIPSITRAC box is replaced within the GCv3 vehicle.
6.2.2.2 Procedure summary
The GIPSITRAC equipment must be warmed up for at least one hour before the calibration is performed. A calibration bench is prepared beside the GCv3 vehicle, which is flat and level. The GIPSITRAC box is removed from the GCv3 vehicle, leaving only the serial data cable to the GCv3 computer and power still attached, and it is placed upon the calibration bench in a specific orientation. GCv3 software within the vehicle is utilised to conduct the calibration. The software directs the placement and
142 Operational Procedures
orientation of the GIPSITRAC box to be changed three additional times while recording the outputs from the accelerometers.
By measuring the outputs from the longitudinal and transverse accelerometers during the different orientations of the GIPSITRAC box, the GCv3 software is able to determine the sensitivity constants of both accelerometers.
The results of the calibration are recorded on a GIPSITRAC bench calibration field sheet and the GIPSITRAC bench calibration process is repeated two additional times to check the consistency of the results.
The GIPSITRAC bench calibration is now complete, however a GIPSITRAC-controlled 180 calibration and a GIPSITRAC-controlled grade and crossfall validation procedure is undertaken before the GIPSITRAC box is installed back in the GCv3 vehicle.
For an example of a GIPSITRAC bench calibration see section 7.6.1 GCv3 GIPSITRAC start-of-season calibration and validation.
6.2.3 GIPSITRAC controlled 180 calibration
The GIPSITRAC box contains MEMS sensors that need to be calibrated in order that the sensor outputs can be correctly and consistently interpreted.
6.2.3.1 Purpose
The purpose of the GIPSITRAC-controlled 180 calibration is to determine the system offset values for the two gyroscopes and the longitudinal and transverse accelerometers within the GIPSITRAC box (ARRB, 1995). Determining the system offset values enables the GIPSITRAC microprocessor to interpret the gyroscope and accelerometer outputs relative to a zero state. The GIPSITRAC-controlled 180 calibration is undertaken whenever a GIPSITRAC bench calibration has occurred and is an interim step before the GIPSITRAC-controlled grade and crossfall validation procedure.
143 Operational Procedures
6.2.3.2 Procedure summary
A GIPSITRAC bench calibration must have been performed immediately prior to this calibration. The GIPSITRAC equipment must be warmed up for at least one hour before the calibration is undertaken. A calibration bench is prepared beside the GCv3 vehicle, which is flat and level. The GIPSITRAC box is removed from the GCv3 vehicle, leaving only the serial data cable to the GCv3 computer and power still attached, and it is placed upon the calibration bench in a specific orientation. GCv3 software within the vehicle is utilised to conduct the calibration. The outputs from the two gyroscopes and the longitudinal and transverse accelerometers are measured. The software directs the placement and orientation of the GIPSITRAC box to be rotated 180 degrees in the horizontal plane. The outputs from the two gyroscopes and the longitudinal and transverse accelerometers are recorded a second time.
By measuring the outputs from the two gyroscopes and the longitudinal and transverse accelerometers before and after rotating the GIPSITRAC box in the horizontal direction, the GCv3 software is able to determine the system offset values for the two gyroscopes and the longitudinal and transverse accelerometers.
The results of the calibration are recorded on a GIPSITRAC bench calibration field sheet, and the GIPSITRAC-controlled 180 calibration process is repeated two additional times to check the consistency of the results.
The GIPSITRAC-controlled 180 calibration is now complete, however a GIPSITRAC controlled grade and crossfall validation procedure needs still to be undertaken before the GIPSITRAC box is installed back in the GCv3 vehicle.
For an example of a GIPSITRAC controlled 180 calibration see section 7.6.1 GCv3 GIPSITRAC start-of-season calibration and validation.
144 Operational Procedures
6.2.4 GIPSITRAC suspension/tilt calibration
The capability to capture road geometry information is highly valued within the RTA. However, the roll and pitch of the vehicle due to the vehicle’s suspension affects the accuracy of the collected road geometry information. Thus it is necessary to correct for the variation due to the suspension effects. This is accomplished via the GIPSITRAC suspension/tilt calibration procedure.
6.2.4.1 Purpose
The purpose of the GIPSITRAC suspension/tilt calibration is to determine the system offset values for the variation in vehicle chassis orientation due to the effect of gravity upon the vehicle suspension relative to the orientation of the road. Determining the system offset values for the vehicle suspension enables the GIPSICAM processing software, GIPSICALC, to correct the orientation of the vehicle for the effect of the vehicle’s suspension and to therefore output road geometry information representative of the road surface. The GIPSITRAC suspension/tilt calibration is undertaken whenever a modification to the vehicle suspension has occurred.
6.2.3.2 Procedure summary
A GIPSITRAC bench calibration, GIPSITRAC-controlled 180 calibration, GIPSITRAC-controlled grade and crossfall validation, and GIPSITRAC daily 180 calibration must have been performed prior to the GIPSITRAC suspension/tilt calibration.
The GIPSITRAC box must be warmed up for at least one hour before the calibration is undertaken. The calibration must occur on a uniformly flat and level surface, and utilises a set of adjustable ramps or an adjustable hoist that can tilt the vehicle to the left, to the right, to the front, and to the back, at angles of up to 30 degrees from the horizontal.
145 Operational Procedures
During the tilt testing, two crash-test dummies are used to simulate the weight of the driver and the operator in vehicle.
First the vehicle is placed on the ramp and the ramp is tilted to the left at an angle of 1 degree from the horizontal. The angle of the vehicle as measured by GIPSITRAC is then recorded. This process is repeated, increasing the tilt of the ramp to the left by 1 degree each time, and measurements made, until an angle of 30 degrees from the horizontal is reached. The result is 30 measurement pairs relating actual ramp angles (from 1 degree to 30 degrees) and the corresponding GIPSITRAC-measured angles that include the effect of the vehicle weight on its suspension.
Figure 76. Transverse suspension/tilt test
The tilt test is then repeated, this time tilting the vehicle to the right, up to an angle of 30 degrees from the horizontal. This tilt test is then repeated two more times, tilting the
146 Operational Procedures
vehicle to the front and to the rear, each time up to an angle of 25 degrees from the horizontal.
Back in the office, the four sets of measurement pairs are used to create a function that is used in the GIPSICAM processing software to convert between measured GIPSITRAC vehicle orientation (longitudinal and transverse angles) and actual road geometry (grade and crossfall).
Figure 77. Longitudinal suspension/tilt test
6.2.5 GIPSITRAC odometer calibration
The process of ‘dead reckoning’ (DR) requires an accurate method of determining the distance travelled. Sensors used to measure distance need to be calibrated to ensure they provide accurate measurements. The primary distance measuring system in the GCv3 vehicle is calibrated using the GIPSITRAC odometer calibration procedure.
147 Operational Procedures
6.2.5.1 Purpose
The primary distance measuring system in the GCv3 vehicle consists of a steel cog attached to the driveshaft of the vehicle and a rotational speed sensor. The distance travelled is determined by counting the teeth of the steel cog as they move past the rotational speed sensor. However, the unknown quantity is what distance on the ground is represented by each cog tooth that passes the sensor. This distance, referred to as the fine-odometer count distance, and a generalised distance where a preset number of teeth are counted as one unit, referred to as the coarse-odometer count distance, is determined by driving the vehicle along an accurately measured distance and recording the number of cog teeth corresponding to this known distance. Dividing the known distance travelled by the total number of cog teeth that passed the rotation speed sensor while travelling the known distance gives the distance for each cog tooth, which is used to specify the fine-odometer count distance and the course-odometer count distance. This GIPSITRAC odometer calibration is conducted at the start of a survey season, and after any change to the vehicle tyres, such as replacement or rotation. The GIPSITRAC odometer calibration is also conducted at the start of any special project where distance is a critical output.
6.2.5.2 Procedure summary
The calibration must occur on a flat stretch of straight road of known length. The distance between the start and end marks of the calibration area must be accurately measured and at least 1000m apart. The start and end points should be clearly marked on the side of the road and/or across the road. A one kilometre long odometer calibration and validation area was established along Napoleon Street in Ramsgate, Sydney.
GCv3 software within the vehicle is utilised to conduct the calibration. The vehicle should be stationary with the rear tyres positioned on the start mark. The exact distance of the calibration area is entered into the odometer calibration software.
148 Operational Procedures
The vehicle is then driven in a straight trajectory along the road towards the end mark, stopping with the rear tyres positioned on the end mark.
The software calculates the fine-odometer count distance by dividing the distance of the calibration line by the number of pulses received from the rotational speed sensor, indicating the number of steel cog teeth that passed the rotational speed sensor. The software then calculates the coarse-odometer count distance by multiplying the fine- odometer count distance by the number of pulses defined as the distance unit.
The calibration is then repeated two more times to check the consistency of the results.
6.2.6 GIPSITRAC daily 180 calibration
6.2.6.1 Purpose
The purpose of the GIPSITRAC daily 180 calibration is to determine the system offset values for the two gyroscopes and the longitudinal and transverse accelerometers within the GIPSITRAC box (ARRB, 1995), which enables the GIPSITRAC microprocessor to interpret the gyroscope and accelerometer outputs relative to a zero state. The GIPSITRAC daily 180 calibration is undertaken each day prior to starting any surveys.
6.2.6.2 Procedure summary
The GIPSITRAC equipment must be warmed up for at least one hour before the calibration is undertaken. The GCv3 vehicle is parked on a uniformly flat area of concrete, asphalt or bitumen. The position and orientation of the passenger side tyres are marked on the ground using chalk. GCv3 software within the vehicle is utilised to conduct the calibration. The outputs from the two gyroscopes and the longitudinal and transverse accelerometers are recorded. The software then directs the driver to “turn the vehicle around 180 degrees” with the driver side tyres positioned and orientated exactly
149 Operational Procedures
with the chalk marks. The outputs from the two gyroscopes and the longitudinal and transverse accelerometers are recorded a second time.
By measuring the outputs from the two gyroscopes and the longitudinal and transverse accelerometers before and after reversing the orientation of the GCv3 vehicle by 180 degrees, the GCv3 software is able to determine the system offset values for the two gyroscopes and the longitudinal and transverse accelerometers.
The GIPSITRAC daily 180 calibration process is repeated two additional times to check the consistency of the results.
6.2.7 GIPSITRAC daily 360 calibration
6.2.7.1 Purpose
The purpose of the GIPSITRAC daily 360 calibration is to determine the sensitivity constants for the two gyroscopes within the GIPSITRAC box (ARRB, 1995). The GIPSITRAC daily 360 calibration is undertaken each day prior to starting any surveys.
6.2.7.2 Procedure summary
The GIPSITRAC equipment must be warmed up for at least one hour before the calibration is undertaken. A large uniformly flat area, such as a parking lot with a looping circuit, is used so that the vehicle can be driven around in large circles. The GCv3 vehicle is parked at the beginning of the loop. GCv3 software within the vehicle is utilised to conduct the calibration. The calibration is commenced and the vehicle is driven around a large circle, stopping at the start of the loop so that the vehicle orientation has rotated horizontally 360 degrees. The outputs from the two gyroscopes are continually recorded.
150 Operational Procedures
By measuring all of the outputs from the two gyroscopes while the GCv3 vehicle drives 360 degrees around the loop, the GCv3 software is able to determine the sensitivity constants for the two gyroscopes.
The GIPSITRAC daily 360 calibration process is repeated an additional time to check the consistency of the results. The GIPSITRAC daily 360 calibration is then repeated a third time, driving in the opposite direction to what was driven the previous two times.
6.3 Validation procedures
Performing the routine calibration procedures ensures the consistency and reproducibility of results. However it is possible for equipment to be configured incorrectly due to user or process error, or equipment that may give inconsistent and unreproducible results due to equipment failure. Thus it is important to be able to verify that all equipment is calibrated correctly, and that all equipment is operating satisfactorily, providing output data to GCv3 system accuracy specifications.
6.3.1 GPS position validation
6.3.1.1 Purpose
The purpose of the GPS position validation is to verify that the equipment is operating according to the manufacturer’s specifications.
6.3.1.2 Procedure summary
A number of coordinated ground stations were installed and accurately surveyed at the Rockdale Works Office in March/April 2006, so that they were in close proximity to the office and workshop. This was done in order to test the GCv3 GPS in different skyview environments,. Note however that any accurately coordinated mark can be used.
151 Operational Procedures
Mark I.D. Easting Northing R.L. Description Accuracy BCJ1 328733.839 6240871.947 2.822 Brass Plug +/- 5mm Den1 328767.922 6240875.428 2.526 Hilti Nail +/- 5mm GP1 328743.87 6240871.74 About 3.0 Gal Post +/- 100mm GP2 328739.85 6240862.64 About 3.0 Gal Post +/- 100mm GP3 328723.65 6240866.31 About 3.0 Gal Post +/- 100mm Table 3. Rockdale coordinated ground stations, referenced in MGA zone 56
The antenna of the GPS to be validated is placed over each of the coordinated ground stations and positions are recorded from the GPS by the data logging device. The recorded coordinates are then compared to the surveyed coordinates, taking into account the environment around the ground stations utilised for the testing and the HDOP calculated from the satellite geometric distribution at the time of the test. If the GPS being tested utilises real time corrections within GCv3 then the real time corrections should be utilised during testing. Accordingly, GPS that are differentially corrected via post-processing in the office should be tested using the corrected positions produced following the GCv3 processing procedures to differentially correct the positions.
6.3.2 GPS + inertial position validation
Absolute GPS positions and relative inertial positions are utilised together to provide position information of the vehicle during the road survey. It is important to have an accurate position of the vehicle throughout the survey as it is this external positioning that is the cornerstone of the integration of the internal vehicle reference system and the camera calibration so that road alignment, road geometry and asset inventories can be related to real world coordinates.
152 Operational Procedures
6.3.2.1 Purpose
The purpose of the GPS + inertial position validation is to verify that the combined technologies are producing an accurate vehicle position during road surveys.
6.3.2.2 Procedure summary
The use of 10cm resolution orthophotography is very convenient for initial testing as it quickly indicates if the combined GPS and inertial data do indeed follow the road carriageway. GIS software such as ESRI ArcGIS can display the road centreline data on top of the 10cm orthophotography. Linking the GIPSICAM images from the survey to the vehicle trajectory data using the ArcGIS hyperlink function allows inspection of the vehicle video to confirm which lane, and where on the road pavement, the vehicle was at that point in time, thus validating the vehicle trajectory data displayed over the orthophotography.
The Bangor Bypass in Southern Sydney was accurately surveyed by RTA surveyors in 2005/2006 for the purposed of validating GIPSICAM alignment and road geometry data (See section 6.3.4 Road geometry validation). Using RTA software, the vehicle trajectory data can be compared against the survey data to verify that the output GCv3 road alignment data meets the specifications.
6.3.3 Camera/lens calibration validation
Calibration of the camera/lens determines the transformation matrix that is used to convert pixel coordinates in a GIPSICAM image to a real world coordinate. This capability to capture the positions of pavement and roadside assets, and to measure/collect width, length, height and area attributes of these assets is crucial, and GIPSICAM was at the forefront of operational road-based MMS technology from the late 1990’s when first deployed. In recent years other operational MMS have
153 Operational Procedures
incorporated some of the advanced asset capture functionality that has been available in GIPSICAM for many years.
6.3.3.1 Purpose
The purpose of the camera/lens calibration validation is to verify that the transformation matrix is indeed correctly converting the pixel coordinates to accurate real world positions.
6.3.3.2 Procedure summary
A stretch of road is required with accurately coordinated reference marks. In addition, accurately measured widths and lengths between the reference marks should also be known. The GCv3 camera/lens calibration validation area is located at the Rockdale Works Office. A straight section of road alongside the workshops provided an excellent test area that was not on a public road, and was convenient to the Rockdale office and workshop just metres away. This section of road is 102m long, with 37 coordinated reference marks that can be used to validate positions, widths, lengths and areas.
Table 4 shows the 33 coordinated references marks that form the testing framework for the 102m Rockdale Test Area. The 33 reference marks define the line marking of an undivided carriageway, consisting of a LHS side of the road, a RHS of the road and the centreline of the road. The Easting (x axis) and Northing (y axis) are a local reference relative to the survey baseline which runs along the LHS of the test area. The Chainage is the distance along the baseline (y axis) to the reference mark. The Offset is the distance across from the baseline (x axis) to the reference mark. The Crown Offset is the distance (x axis) between the LHS or RHS reference mark and the center reference mark. The Position Description is a descriptive name for the reference mark consisting of a prefix of “CH” for chainage, followed by the chainage in meters (no decimal places), then a “-“ character, and finally a character to indicate the side of the road consisting of L (LHS), R (RHS) or C (CENTER).
154 Operational Procedures
The GCv3 vehicle conducts survey runs in both directions of this section of road. The output GIPSICAM data is loaded into the RTA’s AssetLoc software. Using AssetLoc the following information can be collected and compared with the surveyed data to validate the accuracy of the camera/lens calibration: Position of the reference marks. Width between the reference marks. Length between the reference marks. Area defined by three or more reference marks.
For results of a camera/lens calibration validation see section 7.3.1 GCv3 Rockdale test area.
155 Operational Procedures
Local Reference Distance (m) Position Easting Northing Chainage Offset Crown Offset Description 0.6474 -0.0026 0.00 0.65 3.03 CH0-L 3.6769 0.0035 0.00 3.68 CH0-C 6.7842 0.0029 0.00 6.78 3.11 CH0-R 0.6295 9.9945 10.00 0.63 3.04 CH10-L 3.6680 9.9962 10.00 3.67 CH10-C 6.7588 9.9977 10.00 6.76 3.09 CH10-R 0.6451 20.0028 20.00 0.65 3.01 CH20-L 3.6513 20.0019 20.00 3.65 CH20-C 6.7591 19.9987 20.00 6.76 3.11 CH20-R 0.6371 30.0037 30.00 0.64 2.99 CH30-L 3.6221 29.9989 30.00 3.62 CH30-C 6.7802 29.9984 30.00 6.78 3.16 CH30-R 0.6373 40.0031 40.00 0.64 3.01 CH40-L 3.6425 39.9941 40.00 3.64 CH40-C 6.7891 39.9951 40.00 6.79 3.15 CH40-R 0.6444 49.9924 50.00 0.64 2.98 CH50-L 3.6212 50.0032 50.00 3.62 CH50-C 6.7804 50.0050 50.00 6.78 3.16 CH50-R 0.6321 59.9959 60.00 0.63 2.98 CH60-L 3.6126 59.9998 60.00 3.61 CH60-C 6.7990 59.9977 60.00 6.80 3.19 CH60-R 0.6223 69.9982 70.00 0.62 3.01 CH70-L 3.6326 69.9986 70.00 3.63 CH70-C 6.7932 70.0058 70.00 6.79 3.16 CH70-R 0.6141 80.0021 80.00 0.61 3.00 CH80-L 3.6096 80.0002 80.00 3.61 CH80-C 6.7950 80.0027 80.00 6.80 3.19 CH80-R 0.5995 89.9963 90.00 0.60 2.97 CH90-L 3.5668 90.0016 90.00 3.57 CH90-C 6.8373 89.9994 90.00 6.84 3.27 CH90-R 0.5814 101.9996 102.00 0.58 2.96 CH102-L 3.5410 102.0005 102.00 3.54 CH102-C 6.8547 102.0059 102.00 6.85 3.31 CH102-R Table 4. GCv3 Rockdale test area detailed survey data
156 Operational Procedures
6.3.4 Road geometry validation
The estimated accuracy (95% confidence interval) of the road geometry data, as surveyed by GIPSICAM, is (Roads and Traffic Authority, 2006 RGA): Grade: ±1% Crossfall: ±2% Radii: ±20%
6.3.4.1 Purpose
The purpose of the road geometry validation is to verify the accuracy of the road geometry data produced by GCv3.
6.3.4.2 Procedure summary
There are two methods used to validate the road geometry data. The first method is to compare the GCv3 data with accurate survey data. The second method is to validate the reproducibility of the GCv3 data by comparing them against different GCv3 surveys of the same road. The Bangor Bypass in Southern Sydney was accurately surveyed by RTA surveyors in 2005/2006 for the purposed of validating GIPSICAM alignment and road geometry data. Using RTA software, the grade and crossfall data can be compared against the survey data to verify that the output GCv3 road geometry data meets accuracy specifications.
Surveying the same section of road multiple times and comparing the results of the different surveys can validate the reproducibility of the output road geometry data.
For road geometry validation results see section 7.4 GCv3 system georeferenced road geometry.
157 Operational Procedures
6.3.5 Odometer distance validation
6.3.5.1 Purpose
The purpose of the odometer distance validation is to verify the accuracy of the odometers utilised within the GCv3 vehicle.
6.3.5.2 Procedure summary
A 1km long odometer calibration and validation line was established along Napoleon Street in Ramsgate, Sydney, by RTA surveyors. The GCAM software in the GCv3 vehicle can be utilised to measure the length of the Napoleon Street test line, which is then compared to the accurately known distance (of 1km) to validate the accuracy of the odometers.
6.3.6 GIPSITRAC controlled grade and crossfall validation
6.3.6.1 Purpose
The purpose of the GIPSITRAC-controlled grade and crossfall validation is to verify the accuracy of the GIPSITRAC accelerometer calibration.
6.3.6.2 Procedure summary
A GIPSITRAC bench calibration and a GIPSITRAC-controlled 180 calibration must be performed immediately prior to the GIPSITRAC-controlled grade and crossfall validation. A calibration bench is prepared beside the GCv3 vehicle, which is uniformly flat and level. The GIPSITRAC box is removed from the GCv3 vehicle, leaving only the serial data cable to the GCv3 computer and power still attached, and is placed flat
158 Operational Procedures
upon the calibration bench. GCv3 software within the vehicle is utilised to conduct the validation. The outputs from the longitudinal and transverse accelerometers are measured and the calculated grade and crossfall of the calibration bench is compared to the known slope of the calibration bench. This test is repeated three more times, each time rotating the GIPSITRAC 90 degrees in the horizontal plane.
Next, an accurately manufactured wooden ramp at a uniform slope of 5.33% is placed on the calibration bench and the GIPSITRAC is placed flat upon the ramp. The GCAM software is utilised to calculate the grade and crossfall of the ramp, which is compared to the known slope of the ramp. This test is repeated three more times, each time rotating horizontally the GIPSITRAC 90 degrees. The procedure is repeated two more times on ramps of a uniform slope of 10.3% and 15.5%. The results are recorded on a GIPSITRAC bench calibration field sheet.
For an example of the GIPSITRAC controlled grade and crossfall validation see section 7.6.1 GCv3 GIPSITRAC start-of-season calibration and validation.
(See section 6.2.2 GIPSITRAC bench calibration and section 6.2.3 GIPSITRAC controlled 180 calibration).
6.4 Routine procedures
6.4.1 GCv3 GIPSITRAC start-of-season calibration and validation
At the start of each survey season the GIPSITRAC needs to be calibrated and validated (See sections 6.2.2 GIPSITRAC bench calibration, 6.2.3 GIPSITRAC controlled 180 calibration, and 6.3.6 GIPSITRAC controlled grade and crossfall validation). The results are recorded on the GCv3 GIPSITRAC start of season calibration field sheet (See Figure 78). For an example of the GCv3 GIPSITRAC start-of-season calibration and validation see section 7.6.1 GCv3 GIPSITRAC start-of-season calibration and validation.
159 Operational Procedures
GC"3 GIPSITRAC START OF SEASON CALffiRATION "*-'---l..ti
GIPS.l'DAC Rowioll (ABCD): 1 ~ 1 1 ~ 2 1 ~3 1'rDmnt Acc:tltn>aw: ~Att-ome!H:
~ I Cahlndoo2 Calibmiol> 3 G)wscope I : Gylo>"""'f 2: Trmsvtne Acceleromew: ~Accelerometer:
. . No-..ao.. s~Hrt. A CIPSI'I'IIAC Sbrt oiS.... Colillnlioa Sipoff: ------Doto: ____ Figure 78. GCv3 GIPSITRAC start of season calibration field sheet 160 Operational Procedures 6.4.2 GCv3 vehicle daily hours of operation Conducting surveys in metropolitan areas such as Sydney raises the issue of morning and afternoon peak hour traffic that impacts on the efficiency of surveys due to the slower travel times across the road network, and impacts on the quality of the imagery because cars are obscuring the road pavement being videoed. Thus surveys are not conducted during peak hour periods in metropolitan areas. 09:00 Start work 09:00 – 09:30 Start of day checks; GIPSITRAC calibration (180/360) 09:30 – 13:00 AM GIPSICAM survey session 13:00 – 13:30 Lunch 13:30 – 17:00 PM GIPSICAM survey session 17:00 – 17:30 End of day checks; plan for tomorrow; fuel & vehicle maintenance 17:30 Finish work Table 5. Guide to GCv3 hours of operation in metropolitan areas Non-metropolitan areas usually do not have heavy peak hour traffic hence the hours of operation of the vehicle in these areas are generally longer with no need to mitigate the issue of peak hour traffic. 07:00 Start work 07:00 – 07:30 Start of day checks; GIPSITRAC calibration (180/360) 07:30 – 09:30 Early AM GIPSICAM survey session 09:30 – 11:30 Late AM GIPSICAM survey session 11:30 – 12:00 Lunch 12:00 – 14:00 Early PM GIPSICAM survey session 14:00 – 16:00 Late PM GIPSICAM survey session 16:00 – 16:10 End of day checks; plan for tomorrow 16:10 – 16:30 Fuel up vehicle; maintenance 16:30 Finish work Table 6. Guide to GCv3 hours of operation in non-metropolitan areas 161 Operational Procedures Note however that the hours of operation of the GCv3 vehicle is influenced by weather, light conditions, sun angle, sun position, month of the year, mechanical problems and unavoidable dead-running. 6.4.3 GCv3 GIPSICAM survey GIPSICAM surveys are modular in their execution. For example, a survey of the Pacific Highway usually takes four days to complete. The preparation before starting the survey is the “start of project” module. The “start of day” module occurs every day before any survey is commenced. The “pre-run” module is the preparation for starting every survey run. The “post-run” module is the activities that occur after every survey run. The “dead-running” module occurs between all survey runs and is the quality checking of the previous survey run data and getting ready for the next survey run. The “end of day” module is the activities that occur at the end of each day and include the backup of data, preparation of the vehicle for the next day, reporting back to the office regarding the day’s surveys, and securing the vehicle overnight. The list of activities that occur within each survey module are summarised in the GCv3 GIPSICAM survey run-sheet (See Figure 79). 162 Operational Procedures GCv3 GIPSICAM SURVEY RUN-SHEET Dow _.'n'M• 1.111 tWHk O.tt I Drinr I LaCitJ!2[ljL Roa StancP• I Eod Start of project cbeddist: o GJPSICAM system opearional and ftmctionaJ. Port-nm checklist: OCunent CP files copied to ..C: \GC\'3\CP_Files"'. o Vehicle in "park"' \lith bandbrake "on". o Eos:nre eoougb LaCK! Big Disks and "'Run Sheets.. . o Enter «end tink and desc:riptioo" at ibe top of this form. o Bangor calibration nm completed. Yes I No o Ensure \ideo is operating (watch). Yes/ No o Wait 60 seconds (cou:ntdo'wn on PC). Start of day cheddist: o Press "FlO'" key to indicate •'Eod ofRuD'•. a Check brakes, btiDkm and wamiDg tigbts. o Click «Stop Caprure" button & wait 1llltil \ideo refreshes. OGJPSITRAC aDd FOG Figure 79. GCv3 GIPSICAM survey run-sheet 163 Operational Procedures 6.4.4 GCv3 GIPSICAM processing Processing the GIPSICAM data is also undertaken in a modular fashion, as road surveys may have been conducted via multiple survey runs and possibly over multiple days. The “run pre-processing quality check” module is undertaken for every survey run, and consists of confirmation that the survey data has been received and that initial quality checks have been performed. The “run processing” module is undertaken for every survey run, and consists of integrating, adjusting and processing the data, creating all output data for the section of road. The “road post-processing” module is undertaken when all parts of the road have been surveyed and data processed, and consists of combining the results of the various survey runs to produce the data for the entire stretch of road and checking the quality and accuracy of the output data. The “road back up” module is undertaken after each road’s data have been post-processed and checked for quality, and consists of archiving the output GIPSICAM data. A partial list of activities that occur within each processing module are summarised in the GCv3 GIPSICAM processing check-sheet (See Figure 80). 164 Operational Procedures GCv3 GIPSICAM PROCESSING CHECK-SHEET Don _.nnioe. UO Dav of tbe \\'ttk I Date I Timo~l Operator Drinr I LaCieiiDDID Road Name RoadNumbtr Braud a XIR Rua Stan Lillk Ead LiDk Stan CP DtscriptioD I Eud CP DtscriptioD CIIKk-SIIeet Sip:off Run: Pn-Proc.essio: Quality C:btck o Rtm sbeet received. date: ----- o nom checked. o AU cameras checked. o GPS data provided for day: o SSF (GPS) o OBS (OOPS) o OBS (GPS) Pro-Processing QC sign-off: ______date: ____ Run: Processiac Processing lDitiated Date:-,-.,.,.--,- o SSF and OBS saved to projeas\fy_ \raw d:ata\rover\. .. oaspmorecoo o Coo.teots ofGCV3 (raw data) backed up. o GPS data differeDriaJiy corrected: o real rime o from base station o GPF created. o Project Database created: ==-----··gdb o already e:cisted o CWrent Roadl.oc data imponed.. o CWrent CP files created. o atl file created with missing control poims inserted. o HorizoG:ltal Run geoa:neuy c:reated with ioertial and GPS data. o Bo\l·ditcb scale factor appropriate. o Vertic.al nm geomecry adjusted \li th anchor points. o C:o.ssfall geometty adjusted. o Distaoce betwem jpgs set to: o staodatd ( lOm) o other: ___ o Dtfauh image distance set to: o staodatd ( 15m) o other: ___ o Coo.trol points adjusted. o Jpg tablei created. o Jpgs caprured, all images pru.em (no gaps). 0 Jpgs backed up o This is not the finalto:,.,_.,-,;:,-,-===.,-;--::-c=--;:-=....,.-=====, nm in processing for road. Section be.iO\\' immriooa.Uy left blank.. Processing sign-off: ______date : ____ Road: Port-Proc~moc QC o MDB created aod MDB.zip file created: .mdb o MDB and MDB.zip copied to Jpg directory o Link Jeogdls in Assedoc equal to those in CP fik! u LlUk' \Vi.dlkt iu A!.:\IJO:tk>1. ~wd lU dwlit: iu Rulldl.ltu ...~ /Al~o,._hly w llwiwll'-o:tY o CP coordinates iD Assetloc equaJ to those in Roadbrowser o Compare grade and 005sfaU of oe-w data with pre'ious records o Check image points are in appropriate po.s:itions (road ce.oue.lioe) o MDB andMDB.zip backed up to: ______ Post-Processing QC si,gn~ff: ______date : ____ Road: Back-Up o Standard ruobuion backed-up to DVD and eomed into GJPSICA!\ol disk index o High resohltioo backed-up to DVD a:od entered imo GIPSICAM disk inde:c. as: _•os:: ·-======-- o Rtm doct~Dlmted in GJPSICAM progress stUWDal'Y spreadsheet Back-Up si:;n~ff: ______cbte: ____ Comments: Figure 80. GCv3 GIPSICAM processing check-sheet 165 Testing and Validation CHAPTER 7 TESTING AND VALIDATION 7.1 Introduction Initial testing during the development of GCv3 involved familiarisation and experimentation with different items of equipment and subsystems, to assess their capability and suitability to form part of the GCv3 system. As equipment was selected for use in the GCv3 vehicle and the subsystems were integrated, a continual process of testing was implemented to ensure the interoperability of equipment and subsystems within the GCv3. Finally, after building the GCv3 vehicle and developing software and procedures to collect and process the GCv3 data, the GCv3 system was ready for system testing and output data validation. 7.2 GCv3 system vehicle position and road alignment An important fundamental capability of the GCv3 system is “knowing where the GCv3 vehicle is at any point in time during a survey”. Without this accurate positioning data the rest of the information collected, such as road video and road geometry, become less useful. The use of two independent DGPS receivers, a FOG, and a GIPSITRAC ensure multiple sources of ‘absolute’ and ‘relative’ positioning data is recorded and available for utilisation throughout a GCv3 survey. 7.2.1 Display GCv3 vehicle positions in GIS software A straightforward way to verify if the GCv3 vehicle positions for a particular road or survey run are correct is to load the data into a GIS and ‘drape’ over the top of background maps such as accurately ortho-rectified high resolution photography. A visual assessment of the accuracy of the GCv3 vehicle positions data can then be made. 166 Testing and Validation Figure 81. Comparison of GCv3 vehicle positions with 10cm orthophotography When overlayed on accurate orthophotography the vehicle positions data displays as within the road carriageway, thus indicating that the accuracy of the vehicle positions data is suitable for use in GIS software (See Figure 81). If the road or section of road being examined has accurate survey data available, then this data can also be loaded into a GIS and used as a reference feature class or layer to compare against the GCv3 vehicle position data. Python scripts can be developed to automate the comparison (See Figure 90). When overlayed on accurate road boundary survey data the vehicle positions data displays as within the road boundaries, thus indicating that the accuracy of the vehicle positions data is suitable for use in GIS software. If accurate survey data for the road section being examined is not available then previous GIPSICAM surveys can be loaded into a GIS and used as a reference feature class or layer to compare against the current GCv3 survey in order to visually assess the validity of the GCv3 vehicle position data. Again, Python scripts can be developed to automate the comparison. When overlayed with historical GCv3 vehicle positions data the vehicle positions data correlates with the historical vehicle position data to better then ±4.0m indicating a 167 Testing and Validation suitable level of accuracy and reproducibility for use in information systems within the RTA. 7.2.2 Comparison of GCv3 vehicle positions with Bangor Bypass survey data Software was developed by the RTA to compare GCv3 vehicle position data with accurate survey data of the Bangor Bypass calibration/test area, and to produce a report that includes a summary of GCv3 vehicle positional accuracy. Date: 18/05/2007 28/05/2007 5/06/2007 28/09/2007 21/02/2008 Survey Run: ANM ANN ANO ANP ANQ ANR ANS ANT ANU ANA ANB ANZ ANA ANB ANC AHD μ -0.13 1.06 0.22 0.83 0.46 0.16 -0.55 -0.12 0.01 0.08 0.34 0.08 0.21 -0.09 0.24 AHD 2σ 0.35 0.55 0.56 0.47 0.41 0.41 0.34 0.29 0.44 0.44 0.40 0.36 0.30 0.32 0.34 GDA μ 0.74 0.71 0.77 -0.13 0.15 0.57 0.28 -1.71 0.66 -0.39 -0.14 -0.94 0.71 1.15 0.74 GDA 2σ 0.27 0.25 0.28 0.67 0.39 0.25 0.29 0.76 0.46 0.25 0.32 0.28 0.34 0.42 0.21 Table 7. Difference of single observation (AHD/RTA Lambert94) from survey data A compilation of summarised reports showing the comparison of GCv3 vehicle positions with Bangor Bypass survey data can be see above (See Table 7). The table consists of fifteen survey runs, with three survey runs per day, conducted over five different dates. For each survey run the GCv3 vehicle position data was compared with accurate survey data to determine the difference between the two along the whole length of the Bangor Bypass calibration/test area in both the horizontal plane (referenced using RTA Lambert94 and described in the table above as GDA) and the vertical plane (referenced using AHD71 and described in the table above as AHD). The mean (indicated by the symbol “µ”) and two standard deviations (indicated by the symbol “2σ”) of the differences along the whole length of the Bangor Bypass calibration/test area is then calculated in both the horizontal and vertical planes, and described in the table above as “AHD µ” (mean height difference, referenced in meters), “AHD 2σ” (two standard deviations height difference, referenced in meters), “GDA µ” (mean horizontal difference, referenced in meters) and “GDA 2σ” (two standard deviations horizontal difference, referenced in meters). 168 Testing and Validation Using the mean and two standard deviations of the difference between the vehicle positions and survey positions data presented in Table 7 as an indication of the distribution of the results, the worst case distribution can be utilised as an estimate of accuracy under GPS-friendly conditions. As such, the results from the position validation at the Bangor Bypass calibration/test area show that the GCv3 system is determining height (AHD) along the road pavement to an approximate accuracy of ±2.0m and horizontal positions (RTA Lambert94) along the road pavement to an approximate accuracy of ±2.0m (See Table 7). The accuracy of the vertical positions is influenced by errors in the GPS calculation of height, cumulative errors in the INS calculation of change in height, the movement of the vehicle suspension and the operation (acceleration / deceleration) of the vehicle by the driver. The accuracy of the horizontal positions is influenced by errors in the GPS calculation of 2D position, cumulative errors in the INS calculation of change in 2D position and the operation (driving line / position within the lane) of the vehicle by the driver. 7.3 GCv3 system georeferenced road images The camera/lens calibration information enables the determination of real world positions (and dimensions) of road assets (e.g. signs) within the GCv3 road images. If the camera/lens calibration is incorrect then asset positions (and dimensions) will be incorrect. 7.3.1 GCv3 Rockdale test area The GCv3 Rockdale test area is routinely surveyed using GCv3 and the processed data checked using the RTA’s AssetLoc software to ensure the camera/lens calibration is current. Table 8 and Table 9 show the results from GCv3 surveys of the Rockdale test area on 29/11/2007 and 2/12/2008. 169 Testing and Validation Actual lane width (m) GCv3 survey data lane width (m) 28/11/2007 29/11/2007 2/12/2008 Description Width Prescribed Counter Prescribed Counter Prescribed Counter CH0-L to CH0-C 3.03 N/A 3.1 N/A 3.1 N/A 3.1 CH10-L to CH10-C 3.04 3.1 3.1 3.1 3.1 3.1 3.0 CH20-L to CH20-C 3.01 3.1 3.0 3.1 3.0 3.0 3.0 CH30-L to CH30-C 2.99 3.0 3.1 3.0 3.0 3.0 3.1 CH40-L to CH40-C 3.01 3.1 3.1 3.1 3.1 3.0 3.1 CH50-L to CH50-C 2.98 3.0 3.0 3.1 3.0 3.0 3.0 CH60-L to CH60-C 2.98 3.0 3.0 3.0 3.0 3.1 3.0 CH70-L to CH70-C 3.01 3.1 3.1 3.0 3.1 3.0 3.0 CH80-L to CH80-C 3.00 3.1 3.0 3.1 3.0 3.0 3.0 CH90-L to CH90-C 2.97 3.0 3.0 3.0 3.0 3.0 3.0 CH102-L to CH102-C 2.96 3.0 N/A 3.0 N/A 3.0 N/A Table 8. Lane width validation at GCv3 Rockdale test area Actual segment length (m) GCv3 survey data segment length (m) 28/11/2007 29/11/2007 2/12/2008 Description Length Prescribed Counter Prescribed Counter Prescribed Counter CH0-L to CH10-L 10.00 10.1 9.8 10.1 9.8 10.1 10.1 CH10-L to CH20-L 10.00 9.8 10.2 9.9 10.1 10.1 10.0 CH20-L to CH30-L 10.00 10.0 9.9 10.0 9.8 10.0 10.0 CH30-L to CH40-L 10.00 10.2 10.0 10.2 9.9 10.0 10.2 CH40-L to CH50-L 10.00 10.0 10.1 10.1 10.1 10.1 10.0 CH50-L to CH60-L 10.00 10.1 9.8 10.1 9.9 10.1 10.1 CH60-L to CH70-L 10.00 9.8 10.2 9.8 10.2 10.0 10.0 CH70-L to CH80-L 10.00 10.0 10.0 10.0 10.0 10.1 10.0 CH80-L to CH90-L 10.00 10.1 10.2 10.1 10.1 10.0 10.1 CH90-L to CH102-L 12.00 12.1 11.9 12.2 12.0 12.1 12.1 Table 9. Segment length validation at GCv3 Rockdale test area Using the maximum difference between the accurately measured segment width / length and the vehicle survey data segment width / length as presented in Table 8 and Table 9, we can estimate the accuracy of the GCv3 system in determining width dimensions across the road pavement in front of the vehicle and length dimensions along the road pavement in front of the vehicle. The results from the lane width and segment length validation at the GCv3 Rockdale test area confirm that the GCv3 system is determining width dimensions across the road pavement in front of the vehicle to an accuracy of ±0.1m and length dimensions along the road pavement in front of the vehicle to an accuracy of ±0.2m. The consistent 170 Testing and Validation accuracies of the width and length measurements from different survey runs indicates reproducibility of the results. 7.3.2 Comparison with other data sources To validate the lane widths of processed GCv3 data across the state network would require a source dataset; this does not exist. Accurate lane width data for specific project areas around the state have been determined by surveyors and staff in some regions. However, it is easier to use accurate ortho-rectified high resolution aerial photography to determine lane widths and compare them against the lane width data derived from the GCv3 data. Testing has verified that lane widths from 10cm orthophotography are determined to approximately 20cm accuracy. Position details Lane Width (m) RoadLoc reference Orthophotography GCv3 Road No Section Link Chainage (km) As measured 10m 15m 20m MR139 p01 0010 0.190 3.0 3.1 3.1 3.1 p02 0110 0.200 3.3 3.4 3.4 3.5 c01 0090 1.021 3.1 3.1 3.2 3.2 c02 0160 0.421 3.4 3.6 3.4 3.3 MR574 p01 0120 0.250 3.1 3.2 3.1 3.1 p02 1060 0.069 3.2 3.1 3.2 3.3 c01 0120 0.160 3.0 2.9 2.9 2.9 c02 1080 0.100 3.1 3.2 3.1 3.3 MR599 p01 0010 0.070 3.2 3.2 3.3 3.4 p02 0070 0.170 3.2 3.2 3.4 3.4 c01 0055 0.120 4.1 4.0 4.1 4.2 c02 0120 0.160 3.2 3.1 3.2 3.2 MR637 p01 0020 0.220 3.1 3.1 3.1 3.1 c01 0020 0.870 3.0 2.9 3.1 3.0 Table 10. GCv3 lane widths (Nov 2008) vs orthophotography lane widths Using the difference between the orthophotography measured lane width and the vehicle survey data lane width as presented in Table 10, an average difference was determined of 0.079m and a standard deviation of the population sample of 0.068m. 171 Testing and Validation The results show a correlation between the GCv3 lane widths and the lane widths measured from 10cm orthophotography, thus indicating that a comparison of GCv3 measured lane widths and lane widths measured from 10cm orthophotography is a suitable validation option. 7.4 GCv3 system georeferenced road geometry Road geometry data is an important RTA corporate dataset that is used by many staff, such as road designers, maintenance planners and road safety officers. It is important that the road geometry data produced using GCv3 meets the accuracy requirements of the RTA. 7.4.1 Comparison of GCv3 road geometry data with Bangor Bypass survey data Software was developed by the RTA to compare GCv3 road geometry data with accurate survey data of the Bangor Bypass calibration/test area and to produce a report that includes a summary of GCv3 road geometry data accuracy. Date: 18/05/2007 28/05/2007 5/06/2007 28/09/2007 21/02/2008 Survey Run: ANM ANN ANO ANP ANQ ANR ANS ANT ANU ANA ANB ANZ ANA ANB ANC Grade μ -0.01 0.03 0.02 -0.04 -0.03 -0.04 0.00 -0.01 0.01 -0.03 -0.02 -0.02 -0.02 -0.01 0.00 Grade 2σ 0.30 0.32 0.29 0.29 0.37 0.36 0.42 0.44 0.32 0.54 0.43 0.43 0.28 0.28 0.27 Crossfall μ 0.40 0.43 0.45 0.56 0.54 0.55 0.57 0.82 0.51 0.64 0.65 0.70 0.60 0.60 0.64 Crossfall 2σ 0.40 0.42 0.41 0.40 0.47 0.41 0.41 1.49 0.44 0.45 0.45 0.44 0.42 0.43 0.44 Table 11. Difference of single observation (grade/crossfall as %) from survey data A compilation of summarised reports showing the comparison of GCv3 road geometry data with Bangor Bypass survey data can be see above (See Table 11). The table consists of fifteen survey runs, with three survey runs per day, conducted over five different dates. For each survey run the GCv3 road geometry data was compared with accurate survey data to determine the difference between the two along the whole length of the Bangor Bypass calibration/test area for grade (slope measured as %) and crossfall (cross-slope measured as %). The mean (indicated by the symbol “µ”) and two standard 172 Testing and Validation deviations (indicated by the symbol “2σ”) of the differences along the whole length of the Bangor Bypass calibration/test area is then calculated for grade and crossfall, and described in the table above as “Grade µ” (mean grade difference, measured as %), “Grade 2σ” (two standard deviations grade difference, measured as %), “Crossfall µ” (mean crossfall difference, measured as %) and “Crossfall 2σ” (two standard deviations crossfall difference, measured as %). The results from the road geometry validation at the Bangor Bypass calibration/test area show that the GCv3 system is determining grade (slope) along the road pavement to an accuracy better than ±1.0% and crossfall (cross-slope) along the road pavement to an accuracy better than ±2.0%. The consistent average and standard deviation of the measured grade and crossfall data from different survey runs indicates reproducibility of the results. 7.5 GCv3 system road image quality Image quality is an extremely difficult output to test as different levels of quality can be demanded by different applications. However, in general the images should be clear, bright, in-focus, correctly exposed, suitable subject composition, contain no artefacts from the light conditions or system limitations, no motion blur, no vehicle parts visible in the image frame, no reflections on the camera glass, no sun visible in the image frame, and no shadows across the pavement in front of the vehicle. Additionally, while some of these aspects of quality are hardware-related, and some are configuration- related, the others are procedural-related issues that will occur occasionally due to operational requirements such as surveying in the direction towards the sun and surveying in certain directions at certain times of the day that are not optimal. 173 Testing and Validation 7.5.1 Optimal road image quality The overall performance of the video hardware, lenses, and their configuration settings is, in general, very good. Good quality road images are typically obtained in a variety of light conditions and operational environments. Figure 82. Example GCv3 high resolution road image 7.5.2 Hardware related road image quality problems A hardware-related road image quality problem does exist with the GCv3 system. The Sony ICX205AK CCD used in the Sony DFW-SX910 video cameras suffers from visual artefacts such as vertical smearing and blooming. This was a known issue that was deemed an acceptable risk due to similar issues with other CCD’s used in alternative video cameras available in 2005. Testing revealed that the occurrence of the 174 Testing and Validation vertical smearing and blooming was infrequent and, though momentarily distracting, did not impact upon the overall quality of the output road image dataset. Figure 83. An example of vertical smear 7.5.3 Configuration-related road image quality problems Initial pre-vehicle-modification testing revealed four configuration issues relating to road image quality: sunlight shining on the camera glass at an angle causing a ‘smoky’ effect, sunlight shining directly on the camera lens causing a visual artefact, light from inside the vehicle which had the effect of causing the camera glass to reflect like a mirror, and sunlight shining in through the camera glass and reflecting off metal objects back onto the camera glass causing the glass to reflect like a mirror. 175 Testing and Validation 7.5.3.1 Sunlight shining on the camera glass at an angle causing a ‘smoky’ effect To mitigate this problem a window shade was placed over the side camera window and the window shade over the front camera turret was extended. However, occasionally the angle of the sun is such that the sunshine is still able to cause this ‘smoky’ effect. Given the infrequent occurrence of this problem it was decided that it did not impact on the overall quality of the output road image dataset. Figure 84. An example of the “smoky” effect 7.5.3.2 Sunlight shining directly on the camera lens causing a visual artefact The addition of a window shade for the side camera window and the extension of the window shade of the front camera turret also helped to address this problem. However, as with the ‘smoky’ effect, it would be impossible to fully mitigate this problem as the 176 Testing and Validation sun sometimes gets low enough in the sky that it will be “looking back in at the camera”. Procedures such as not driving in the direction of the sun help to resolve this issue for the front camera, however the side camera can still suffer from this problem. Given the infrequent occurrence of the issue, particularly when survey procedures are followed, it was decided that it did not impact on the overall quality of the output road image dataset. Figure 85. An example of sunlight shining directly on the camera lens 7.5.3.3 Light from inside the vehicle causing camera glass reflections To mitigate this issue a black curtain was added to the back of the camera shelf. Occasionally though, new staff forget to “close” the camera shelf curtain and this effect can still be seen. Staff training resolves this issue. 177 Testing and Validation 7.5.3.4 Sunlight reflecting off metal objects causing camera glass reflections The sunlight can be reflected off a number of objects within the camera shelf area, but it is the metal base-plate, the camera body and the camera lens that reflect the sunlight the most. To mitigate this issue a shroud of black cloth was added around each camera and the camera base-plate. Figure 86. An example of the mirror effect 7.5.4 Procedural related road image quality problems Procedural-related issues are primarily conducting surveys at inappropriate times of the day, such as late in the afternoon when shadows are long, or at inappropriate times of the year, such as winter when the days are shorter and the sun angle is generally lower thereby causing shadows and poor lighting issues. They are also caused by conducting 178 Testing and Validation surveys in an easterly direction in the morning and a westerly direction in the afternoon, when the sun angle is low and shining directly into the camera lens. The solution is to ensure that the staff are trained in the survey procedures and have an understanding of the issues that cause poor image quality in order to avoid them occurring. 7.6 GCv3 GIPSITRAC calibration and validation The GIPSITRAC box contains MEMS sensors that need to be calibrated so that the sensor outputs can be correctly and consistently interpreted. The calibration procedures determine the sensitivity constants and the system offset values for the two gyroscopes and for the longitudinal and transverse accelerometers within the GIPSITRAC box (ARRB, 1995). 7.6.1 GCv3 GIPSITRAC start-of-season calibration and validation At the start of each survey season the GIPSITRAC MEMS sensors are calibrated using the GIPSITRAC bench calibration procedure (See section 6.2.2 GIPSITRAC bench calibration) and the GIPSITRAC controlled 180 calibration procedure (See section 6.2.3 GIPSITRAC controlled 180 calibration). The calibration of the GIPSITRAC MEMS accelerometers are then validated using the GIPSITRAC controlled grade and crossfall validation procedure (See section 6.3.6 GIPSITRAC controlled grade and crossfall validation). The results are recorded on the GCv3 GIPSITRAC start of season calibration field sheet (See Figure 87). 179 Testing and Validation GCv3 GIPSITRAC START OF SEASON CALIBRATION Document \'ersion LOS GIPSITRAC Box Number: Date of Calibration: GIPSITRAC ON > 12 hours: Officer Name: Calibration Bench leveJJStable: Officer Position: 1. Bench Calibration (ARRB New Package Calibration): GIPSITRAC Rotation (ABCD): Transverse Accelerometer: Longitudinal Accelerometer: 2. Controlled 180 Calibration (Static Calibration I ARRB 180 Calibration): Calibration I Calibration?- Calibration 3 Gyroscope I: -7· SS -11 · 1 ~ -7 ·'+4- Gyroscope 2: -30G·o7 -3oS ·S~ -JOS•'f7 Transverse Accelerometer: -0·8S -0·71 -o·63 Longitudinal Accelerometer: S· 8~ S• 7~ 5'•68 3. Controlled Grade and Crossfall Validation (Tilt T est I Spring Calibration): Nominal 0% slope. Actual slope (digital spirit Ievell: (Grade: O•l % lfX-fall: 0•0 % ) Handle towards (0°) Grade (longitudinal): 0•11 % X-fall (transverse): -o·os % Han dle away ( 180°) Grade (longitudinal): -o·l"- % X-fall (transverse): 0•06 % Han dle to left (90°) X-fall (longitudinal): 0·0") % Grade (transverse): 0 •18 % Handle to right (270°) X-fall (longitudinal): -o·o't % Grade (transverse): -o·Jtf- % Nominal 5% slope. Actual slope (di!:ital spirit level): !Grade: 5·ll- %l(X-fall: 0•1 % ) Handle towards (0°) Grade (longitudinal): -S·'+I % X-fall (transverse): 0•06 % Handle away ( 180°) Grade (longitudinal): S· '+~ % X-fall (transverse): -o ·ol % Handle to left (90°) X-fall (longitudinal): -o · o6 % Grade (transverse): - s·<+O % Handle to right (270°) X-fall (longitudinal): O•d6 % Grade (transverse): s · '+a % NominallO% slope. Actual slope (dieital spirit level): (Grade: 10·3 %)(X-fall: 0 · ~ % ) Handle towards (0°) Grade (longitudinal): - 1 o • 3 3 % X-fall (transverse): o.:J. 7 % Handle away (180°) Grade (longitudinal): I 0 •3 3 % X-fall (transverse): -o-:2.S % Handle to left (90°) X-fall (longitudinal): - 0. z.~ % Grade (transverse): -Jo ·:J.q % Handle to right (270°) X-fall (longitudinal): 0 ·;1. 4- % Grade (transverse): 10 •38 % Nominal IS% slope. Actual slope (dinital spirit level): (Grade: 15'•3 %lfX-fall: o•.;t % ) Handle towards (0°) Grade (longitudinal): _, :>•36 % X-fall (transverse): 0 •30 % Handle away ( 180°) Grade (longitudinal): /S•3S' % X-fall (transverse): -0 ·7:.4- % Handle to left (90°) X-fall (longitudinal): -0·2.'+- % Grade (transverse): -ts · 36 % Handle to right (270°) X-fall (longitudinal): 0 ·<. s % Grade (transverse): IS ·3 7 % Comments: 1);~,=\= GlPSlTRACStartofS e ason C alibratiou S ignoff: "V~ ~ Date: ~o/'1/:J..oo7 GCv3 GJPSITRAC Start ofS easo11 Ca/ibrario11 procedures created by De1111is E11trike11 Figure 87. GCv3 GIPSITRAC start of season calibration results 180 Testing and Validation The GIPSITRAC bench calibration procedure is repeated three times and the determined sensitivity constants for the longitudinal and transverse accelerometers are recorded in section “1. Bench Calibration (ARRB New Package Calibration)” on the GCv3 GIPSITRAC start of season calibration field sheet (See Figure 87). The three measured sensitivity constants should be reproducible to a precision of ±0.01 for each accelerometer respectively. The GIPSITRAC controlled 180 calibration procedure is repeated three times and the determined system offset values for the two gyroscopes and the longitudinal and transverse accelerometers are recorded in section “2. Controlled 180 Calibration (Static Calibration / ARRB 180 Calibration)” on the GCv3 GIPSITRAC start of season calibration field sheet (See Figure 87). The three measured system offset values should be reproducible to a precision of ±1.0 for each gyroscope respectively and ±0.3 for each accelerometer respectively. The GIPSITRAC controlled grade and crossfall validation procedure measures four known slopes and x-slopes, each in four horizontal orientations (relative to the upper surface of the calibration bench) and the calculated grade and crossfall from the longitudinal and transverse accelerometers are recorded in section “3. Controlled Grade and Crossfall Validation (Tilt Test / Spring Calibration)” on the GCv3 GIPSITRAC start of season calibration field sheet (See Figure 87). The measured grades and crossfalls should all have accuracy to within ±0.1% when compared with the known slopes and x-slopes. 7.6.2 GIPSITRAC daily 180 calibration and daily 360 calibration At the start of each survey day the GIPSITRAC MEMS sensors are calibrated using the GIPSITRAC daily 180 calibration procedure (See section 6.2.6 GIPSITRAC daily 180 calibration) and the GIPSITRAC daily 360 calibration procedure (See section 6.2.7 GIPSITRAC daily 360 calibration), as recommended by the GIPSITRAC operational manual (ARRB, 1995). 181 Testing and Validation The GIPSITRAC daily 180 calibration determines the system offset values for the two gyroscopes and the longitudinal and transverse accelerometers within the GIPSITRAC box (ARRB, 1995). The GIPSITRAC daily 360 calibration determines the sensitivity constants for the two gyroscopes within the GIPSITRAC box (ARRB, 1995). The results of the GIPSITRAC daily 180 calibration and the GIPSITRAC daily 360 calibration are saved in the “GipsiCamConstants.txt” file, and also appended to the “GCamConstantsHistory.txt” file. See Table 12 for a sample of GIPSITRAC gyro and accelerometer calibration values. The GipsiCamConstants.txt file is the source ASCII data used as the .INE file header record (See section 4.3.3.1.3.10 GCAM .INE file). The calibration details saved in the GCamConstantsHistory.txt consist of: Date Time Gyro 1 Zero Offset Gyro 1 Sensitivity Gyro2 Zero Offset Gyro2 Sensitivity Transverse Accelerometer Zero Offset Transverse Accelerometer Sensitivity Longitudinal Accelerometer Zero Offset Longitudinal Accelerometer Sensitivity Course Odo Fine Count Pitch Roll 182 Testing and Validation 7.6.3 Analysis of GIPSITRAC gyroscope and accelerometer calibration values The performance of the GIPSITRAC MEMS sensors is affected by the ambient temperature of their operating environment, thus the need to ensure that the GIPSITRAC was ‘warmed up’ to operating temperature and calibrated using the daily GIPSITRAC calibration procedures before commencing surveys (ARRB, 1995). However, now that the rear of the GCv3 vehicle is cooled via an independent air- conditioner (See section 3.3.3 Air conditioning) then do we need to continue the daily GIPSITRAC calibrations if the ambient temperature in the rear of the GCv3 vehicle is controlled its own air conditioner? Table 12 shows the GIPSITRAC gyro and accelerometer calibration values over a period of two working weeks from Monday 6/11/2006 to Friday 17/11/2006, during operational GIPSICAM surveys and following the GCv3 calibration and validation procedures described in CHAPTER 6 OPERATIONAL PROCEDURES. At the bottom of Table 12 I have calculated an average and a standard deviation for each of the eight columns of calibration values. Note that the different GIPSITRAC calibration procedures each calibrate a different subset of sensors and variables, so a more accurate way to analyse the data would be to extract the data specifically resulting from each GIPSITRAC calibration procedure and calculate the statistics separately. However, for the purpose of this analysis the method undertaken is acceptable. The accelerometer sensitivity values are constant as they are only calibrated during a GIPSITRAC bench calibration (See section 6.2.2 GIPSITRAC bench calibration) which is at the start of a survey season or when the GIPSITRAC is removed and replaced within the GCv3 vehicle. The gyroscope and accelerometer zero offset values are determined during the GIPSITRAC daily 180 calibration (See section 6.2.6 GIPSITRAC daily 180 calibration), while the gyroscope sensitivity values are determined for the GIPSITRAC daily 360 calibration (See section 6.2.7 GIPSITRAC daily 360 calibration). 183 Testing and Validation Gyroscopes Accelerometers System Gyro 1 Gyro 2 Transverse Longitudinal Date Time Zero Offset Sensitivity Zero Offset Sensitivity Zero Offset Sensitivity Zero Offset Sensitivity 6/11/2006 11:16:26 -112.3786 -2474.1840 -142.6462 2509.3743 -7.3929 -172.1713 26.8081 -107.9152 6/11/2006 11:19:58 -112.3786 -2503.9520 -142.6462 2533.0286 -7.3929 -172.1713 26.8081 -107.9152 6/11/2006 11:22:06 -112.3786 -2505.5827 -142.6462 2534.3728 -7.3929 -172.1713 26.8081 -107.9152 6/11/2006 11:24:04 -112.3786 -2515.1057 -142.6462 2543.3510 -7.3929 -172.1713 26.8081 -107.9152 6/11/2006 11:26:06 -112.3786 -2518.3423 -142.6462 2544.3624 -7.3929 -172.1713 26.8081 -107.9152 7/11/2006 07:56:51 -113.2648 -2518.3423 -142.5405 2544.3624 -6.7976 -172.1713 27.0038 -107.9152 7/11/2006 07:59:40 -113.2648 -2515.5631 -142.5405 2548.8461 -6.7976 -172.1713 27.0038 -107.9152 7/11/2006 08:01:32 -113.2648 -2518.4803 -142.5405 2544.8324 -6.7976 -172.1713 27.0038 -107.9152 8/11/2006 08:14:24 -108.4443 -2518.4803 -142.9548 2544.8324 -6.8943 -172.1713 28.0876 -107.9152 8/11/2006 08:19:26 -108.4443 -2518.0282 -142.9548 2546.7860 -6.8943 -172.1713 28.0876 -107.9152 8/11/2006 08:21:16 -108.4443 -2524.2200 -142.9548 2547.9483 -6.8943 -172.1713 28.0876 -107.9152 9/11/2006 07:24:14 -112.8810 -2524.2200 -140.9460 2547.9483 -9.2665 -172.1713 28.3845 -107.9152 9/11/2006 07:26:25 -112.8810 -2415.5324 -140.9460 2552.7268 -9.2665 -172.1713 28.3845 -107.9152 9/11/2006 07:30:33 -112.8810 -2620.2698 -140.9460 2551.0928 -9.2665 -172.1713 28.3845 -107.9152 9/11/2006 07:32:45 -112.8810 -2609.8481 -140.9460 2550.6398 -9.2665 -172.1713 28.3845 -107.9152 10/11/2006 08:23:21 -112.3229 -2609.8481 -123.4529 2550.6398 -7.5976 -172.1713 30.4710 -107.9152 10/11/2006 08:28:57 -112.3229 -2465.2886 -123.4529 2540.0250 -7.5976 -172.1713 30.4710 -107.9152 10/11/2006 08:30:49 -112.3229 -2463.6853 -123.4529 2541.2312 -7.5976 -172.1713 30.4710 -107.9152 10/11/2006 08:34:43 -112.3229 -2469.7583 -123.4529 2548.1734 -7.5976 -172.1713 30.4710 -107.9152 13/11/2006 08:51:52 -107.3019 -2469.7583 -171.2748 2548.1734 -6.7229 -172.1713 27.3448 -107.9152 13/11/2006 08:55:05 -107.3095 -2469.7583 -171.3686 2548.1734 -5.7673 -172.1713 28.0264 -107.9152 13/11/2006 08:59:20 -107.3095 -2528.2308 -171.3686 2540.3333 -5.7673 -172.1713 28.0264 -107.9152 13/11/2006 09:00:53 -107.3095 -2531.6270 -171.3686 2545.2912 -5.7673 -172.1713 28.0264 -107.9152 13/11/2006 09:02:13 -107.3095 -2526.3223 -171.3686 2540.1550 -5.7673 -172.1713 28.0264 -107.9152 14/11/2006 07:09:14 -122.0062 -2526.3223 -109.7814 2540.1550 -4.4367 -172.1713 28.9862 -107.9152 14/11/2006 07:20:16 -122.0276 -2526.3223 -109.5414 2540.1550 -3.8324 -172.1713 29.4995 -107.9152 14/11/2006 07:23:52 -122.0276 -2463.2666 -109.5414 2551.5954 -3.8324 -172.1713 29.4995 -107.9152 14/11/2006 07:25:29 -122.0276 -2460.0748 -109.5414 2548.8968 -3.8324 -172.1713 29.4995 -107.9152 14/11/2006 07:31:13 -122.0276 -2487.8186 -109.5414 2558.5061 -3.8324 -172.1713 29.4995 -107.9152 14/11/2006 07:33:33 -122.0276 -2482.3671 -109.5414 2554.2322 -3.8324 -172.1713 29.4995 -107.9152 15/11/2006 06:55:28 -116.4295 -2482.3671 -120.1233 2554.2322 -6.0219 -172.1713 30.0052 -107.9152 15/11/2006 07:00:17 -116.5248 -2482.3671 -120.0886 2554.2322 -6.9324 -172.1713 26.8795 -107.9152 15/11/2006 07:06:06 -116.5248 -2490.4961 -120.0886 2565.4828 -6.9324 -172.1713 26.8795 -107.9152 15/11/2006 07:09:31 -116.5248 -2501.0046 -120.0886 2567.4518 -6.9324 -172.1713 26.8795 -107.9152 16/11/2006 06:49:09 -111.3619 -2501.0046 -136.6071 2567.4518 -6.6867 -172.1713 27.1776 -107.9152 16/11/2006 06:55:15 -111.3995 -2501.0046 -136.4857 2567.4518 -12.3152 -172.1713 28.6338 -107.9152 16/11/2006 07:14:50 -111.3995 -2490.7963 -136.4857 2563.4487 -12.3152 -172.1713 28.6338 -107.9152 16/11/2006 07:22:20 -111.3995 -2489.8677 -136.4857 2567.5239 -12.3152 -172.1713 28.6338 -107.9152 16/11/2006 07:24:13 -111.3995 -2493.7123 -136.4857 2556.6480 -12.3152 -172.1713 28.6338 -107.9152 16/11/2006 07:25:48 -111.3995 -2494.1123 -136.4857 2555.3195 -12.3152 -172.1713 28.6338 -107.9152 17/11/2006 07:47:15 -114.5671 -2494.1123 -130.1348 2555.3195 -5.9886 -172.1713 27.5243 -107.9152 17/11/2006 07:51:12 -114.5814 -2494.1123 -130.3771 2555.3195 -7.0414 -172.1713 30.4367 -107.9152 17/11/2006 07:55:55 -114.5814 -2456.4737 -130.3771 2540.6204 -7.0414 -172.1713 30.4367 -107.9152 17/11/2006 07:57:44 -114.5814 -2459.1784 -130.3771 2540.4299 -7.0414 -172.1713 30.4367 -107.9152 17/11/2006 08:00:13 -114.5814 -2463.1703 -130.3771 2561.9857 -7.0414 -172.1713 30.4367 -107.9152 17/11/2006 08:02:39 -114.5814 -2466.4156 -130.3771 2566.9743 -7.0414 -172.1713 30.4367 -107.9152 Average: -113.4426 -2500.8869 -135.2825 2549.5681 -7.2862 -172.1713 28.5080 -107.9152 STD Deviation: 4.1829 39.2631 16.8915 11.0245 2.2380 0.0000 1.3020 0.0000 Table 12. GIPSITRAC gyro and accelerometer calibration values over two weeks Looking at the standard deviations we can see that there is still a slight variation in the calibration values each day. My assumption for this is that the temperature in the rear of the vehicle, while having a separate air-condition now, does not have an accurate thermostat controlled environment, and the access to the rear of the GCv3 vehicle is via 184 Testing and Validation the cargo door or the rear door which lets the cool air escape from the rear of the vehicle. Suggestions for future research include the addition of an accurate thermostat control for the rear of the GCv3 vehicle; the replacement of the GIPSITRAC MEMS sensors for sensors that do not require calibration; the replacement of GIPSITRAC for an INS that does not require calibration but is still able to provide road geometry data; and the further testing of the actual significance of the variation of the daily calibration values to determine if the daily calibrations are really required or if an average value could be used. 185 Application of MMS Technology CHAPTER 8 APPLICATION OF MMS TECHNOLOGY 8.1 Introduction This chapter describes the application of Mobile Mapping Technology within a Roads and Traffic Authority to assist individual business units, and the organisation as a whole, to achieve operational objectives. An overview of the different datasets produced from the GIPSICAM technology is presented and some specific examples of applying GIPSICAM Technology are described, showing how such utilisation leads to the provision of better services to clients in terms of efficiency and road user safety. 8.2 Routine applications of the GIPSICAM data within the RTA “The RTA employs about 6,900 staff in more than 180 offices throughout NSW, including 129 motor registries” (Roads and Traffic Authority, 2007). There are currently 1,800 RTA staff members that use GIPSICAM data on a regular or semi-regular basis to perform their duties. Excluding motor registry staff members, who deal more with licensing and registration, it becomes apparent that a very high percentage of RTA staff use GIPSICAM data. The “Regional Operations and Engineering Services”, “Major Infrastructure” and “Network Management” directorates, as well as the “Road Safety” branch, are the biggest users. 186 Application of MMS Technology 8.2.1 Road centreline vector data The GIPSICAM road centreline data is used as a primary data source in the maintenance of the NSW classified roads dataset. So how is the GIPSICAM road centreline data produced? GPS/DGPS data is susceptible to errors caused by surrounding buildings, terrain and trees. GPS/DGPS signal can be lost or degraded underneath tree canopies, in tunnels, next to very tall buildings and road cuttings, and in mountainous areas. Errors such as multipath are common in areas of tall buildings, bridges, other structures and cliffs. The number of available satellites and the satellite geometry can also affect the accuracy of the GPS/DGPS pseudo-range position results. Thus, while DGPS data can be accurate to less than one metre (2D) in favourable locations with good visibility and geometry conditions, DGPS can still be affected by signal loss/degradation, multipath and bad satellite geometries, which will reduce the accuracy of DGPS data (e.g. Leica Geosystems Inc, 1999). INS systems can determine relative position, based on the previous position plus information measured from the INS sensors and an odometer. Errors that occur are cumulative and thus, while an INS system may be accurate for short periods of time, the INS-determined positions will become less and less accurate with time if relied upon as the sole positioning technology (e.g. Ford et al, 2004). As already mentioned, DGPS is susceptible to errors from signal loss/degradation, multipath or bad satellite geometry, however DGPS using pseudo-range measurements can provide discrete positions to sub-metre accuracy. On the other hand, INS data is not affected by signal loss/degradation, multipath or satellite geometry, and is thus continually available, although it suffers from the aforementioned cumulative errors. However, if DGPS and INS are combined then advantage can be taken of sub-metre accurate discrete positions supplemented by continuously available INS data (e.g. Li et al, 2005). 187 Application of MMS Technology The RTA has developed a Least Squares adjustment program that combines the DGPS data from the DGPS/GPS receivers and the INS data from GIPSITRAC and the FOG to generate accurate GCv3 vehicle position and attitude information (See Figure 12). The GCv3 vehicle surveys “undivided carriageway” in both directions, always driving in the through-lane closest to the road centreline. Thus, if one averages the vehicle position data for both directions then the result is an approximation of the location of the true road centreline. The GCv3 vehicle surveys “divided carriageway”, such as on a freeway or a road with a concrete median down the centre of the road, in the normal direction of traffic flow, but always driving in the centre through-lane if there are three lanes, or always driving in the left lane through-lane if there is only two lanes, or always driving in the through- lane if there is only one lane. Thus the vehicle position data is an approximation of the actual carriageway centreline, with the exception of two lane roads where an offset is applied to the vehicle position data to generate an approximation of the actual carriageway centreline. Approximating the road centreline as described above is a much simpler method of accurately determining the location of the road centreline, as opposed to methods that try to extract the centreline line markings from georeferenced images. In areas of good DGPS coverage, the GIPSICAM-derived centreline data is accurate to better than one metre. In more challenging environments, such as in the CBD with tall buildings, through a state forest with tree canopy above the road, or through any of Sydney’s tunnels, then the centreline data accuracy will be reduced. Based on comparisons with other datasets, such as 10cm orthophotography and field survey data, RTA staff estimate the horizontal accuracy of the state-wide GIPSICAM-derived vehicle position dataset to be sub-two-metres (95% confidence interval). The use of the OmniSTAR HP DGPS correction service would potentially increase this accuracy to the decimetre level (OmniSTAR, 2011), while utilising real time kinematic (RTK) technology would potentially increase the accuracy to the centimetre level (Wikipedia, 2011a). 188 Application of MMS Technology Pre-2005, the NSW road centreline data maintained by the NSW Department of Lands was acknowledged as being inaccurate and a project was established with the aim of increasing the accuracy of that dataset (Department of Lands, 2005). The RTA and the NSW Department of Lands came to an agreement that the RTA would build and maintain an accurate NSW classified roads centreline dataset that would be provided to the NSW Department of Lands and other state government departments. The RTA created the new dataset with the alignment of the state roads based primarily on GIPSICAM centreline data, supplemented with high-resolution orthophotography and road design data. Newly built and/or gazetted state roads and roads with altered alignments are surveyed by the GCv3 vehicle, and the new centreline data is used to maintain the NSW classified roads centreline dataset. The NSW classified roads centreline dataset is available in the RTA via ESRI SDE and can be displayed in ESRI ArcMap with other RTA corporate data. It is also used as the framework for the spatial component of the RTA’s Road Asset Maintenance System. 8.2.2 Georeferenced road images As the well-known proverb goes, “a picture is worth a thousand words”. This is certainly true when it comes to terrestrial road images, as an image typically contains much information. To be able to look at an image and instantly “see” what asset items are present, and what their condition was at the time the image was captured, has proved very useful within the RTA. Add to this the capability to extract asset position, dimension, condition and attribute information and one begins to appreciate the potential of an MMS (See Figure 88). 189 Application of MMS Technology Figure 88. GCv3 georeferenced road image By far the most common application of MMS road images is, as mentioned, for the visual inspection of the road condition/characteristics. This statement is confirmed by Meers (2007) who states “the images are most commonly used for simply ‘visualising’ a section of road - whether it is to confirm one's mental image of the section, or create a new understanding of the road and it's environment for those that are not familiar with it, or to pick exact locations of certain features in relation to others”. The advantage of using the road images as a visual data source is that in most cases the RTA staff members do not need to travel to a specific location or worksite to undertake their work. “In terms of occupational health and safety (OH&S), one of the major benefits of utilising the road images is the level of protection given to RTA staff by allowing them to perform their duties in a safe environment rather than having to actually walk out onto the road” stated Dunlop (2007). Alternatively, the road image data could be used to prepare for an actual roadside inspection. This saves time and money in unnecessary travel. Avoiding unnecessary travel to a roadside location is 190 Application of MMS Technology particularly important in remote areas such as in the “Western” and “South-West” regions of NSW. In the case of historical data there is no substitute for an image record. Lastly, the ability to capture an asset inventory quickly, easily and accurately without the need to travel onsite is an inexpensive and easily repeatable data collection process. Examples of road image use include: Asset inspection, identification, validation and remaining life determination by asset managers, maintenance planners, designers and engineers. Asset inventory data collection for import into the RTA’s Road Asset Maintenance System or the RTA’s corporate GIS. Asset dimensional characteristics (position, width, height, length, area). Road safety audit and analysis of safety related assets. Critical habitat and environmental corridor determination/inspection by environmental officers. Surveillance of maintenance and construction works by RTA officers. Sharing a “common frame-of-reference”, as in looking at the same road image, when talking to other RTA business units. Scoping of works with stakeholders such as local government and contractors. Verification of segment markings on the side of the road. Inspection of current and historical image data to show the presence/absence or condition of assets and the roadside environment. Visualisation of road geometry and road condition data. Verification and geocoding of crash sites from police reports. SCRIM (Sideways Force Coefficient Routine Investigation Machine) site category determination (See 8.3.8 SCRIM site categories determination). Litigation enquiries involving road condition or the presence/absence of an asset such as a speed sign or a stop sign. 8.2.3 Georeferenced road geometry data “GIPSITRAC provides an accurate and comprehensive record of the geometry of a road” (Roper, 2003). GIPSITRAC is an acronym for “Global and Inertial Positioning 191 Application of MMS Technology System Integration for Tracking Route Alignment and Crossfall” (ARRB, 1995). The GIPSITRAC “box” was first designed and built by ARRB in the early 1990’s and consists of gyroscopes, accelerometers, GPS and a sensor to measure distance. Using data from the gyroscopes, accelerometers and the mechanical odometer, the GIPSICAM system can calculate road geometry information, specifically grade, crossfall, horizontal radius and vertical radius. Other RTA software can derive and display additional information such as horizontal curvature, vertical curvature, advisory speed, transverse (centripetal) acceleration, vertical acceleration, longitudinal acceleration, combined vertical/transverse acceleration, combined longitudinal/transverse acceleration, stop sight distance and “K value” (Roads and Traffic Authority, 2006 RGA). The accuracy of the GIPSITRAC box according to its design specifications (ARRB, 1995) is: Grade: 0.2% Crossfall: 0.2% Curvature: 0.1 radian/km Known limitations of the surveyed GIPSITRAC road geometry data include (Roads and Traffic Authority, 2006 RGA): Geometry data is valid for GIPSITRAC/GIPSICAM vehicle path only. Horizontal and vertical radii can be affected by the skill and experience of the driver. Examples: vehicle entry and exit angle of a curve; oversteering and understeering through corners; vehicle “wander”. Grade can be affected by the skill and experience of the driver. Examples: braking heavily; accelerating heavily. Crossfall can be affected by cross-winds and centrifugal forces acting on the vehicle suspension when going around corners. The limitations affecting the road geometry data are minimised by ensuring the GIPSICAM drivers are skilled drivers, are very experienced at driving the GIPSICAM 192 Application of MMS Technology vehicle and are aware of the situations that may affect the accuracy of the road geometry data. The estimated accuracy (95% confidence interval) of the road geometry data, as surveyed by GIPSICAM, is (Roads and Traffic Authority, 2006 RGA): Grade: ±1% Crossfall: ±2% Radii: ±20% The advantage of collecting road geometry data via an MMS rather than traditional surveying methods is a matter of time and cost. An MMS can survey 80km of road per hour. While the data collected by traditional surveying methods is more accurate, the rate at which traditional survey methods collect data cannot compete with a MMS in terms of throughput. The road geometry data is georeferenced so that it can be displayed and analysed in a GIS. The RTA has developed a computer program that plots and correlates road geometry data, including derived data, for analysis. The georeferenced road images are also linked to the road geometry data for visual inspection, feature interrogation or asset information extraction. The use of the road geometry data within the RTA is increasing rapidly, with the Road Safety Branch finding the data particularly useful for analysing accident data. Designers are also finding the geometry data a useful tool to be used in conjunction with the RTA developed “Brownfields Design Guide”, a road design guide that describes solutions to designing safe roads with limited funds (See Figure 89). 193 Application of MMS Technology Figure 89. A sample plot of GCv3 road geometry data Examples of road geometry data use include: Analysis of accident data by correlating accident sites against road geometry. Pavement rehabilitation projects using the Brownfields Design Guide. Heavy vehicle route planning. Bus route safety analysis. NSW road network safety analysis. Road-water runoff determination. SCRIM site category determination. Litigation enquiries involving road geometry. Historical enquiries involving road geometry. 194 Application of MMS Technology 8.3 Specific applications of the GIPSICAM data within the RTA A short description of a selection of MMS related projects undertaken within the RTA is given below. 8.3.1 Asset data collection The RTA has been conducting an inventory of all lanes data, on all NSW state roads, for the past 18 months using GIPSICAM’s georeferenced road images. The lanes information collected includes primary and second functions, and is comprehensive in terms of detail and state road coverage. The lanes data is imported into the RTA’s Road Asset Maintenance System (RAMS). Road Safety Branch used GIPSICAM’s georeferenced road images to conduct an inventory of “crash related” road assets within rural areas while building their Rural Roads Stereotypical Crash Rates database. Assets collected included speed signs, crash barriers, pavement surface type, road shoulder width, median width, bridges, run-off areas and accesses/driveways. Analysing crash location data against the “crash related” asset data collected from the road images, staff from Road Safety Branch were able to determine parameters that appeared to influence crash rates on rural roads, which yielded a mechanism for determining average crash rates for different stereotypes of rural road (Chee, 2005). According to Tang (2007), “Safer Roads Section is currently undertaking a study to review and update the analysis methods and data within the Rural Roads Stereotypical Crash Rates database”. GIPSICAM data will be used to conduct the update of the “crash related” road assets inventory. The RTA conducts an annual inventory of road features related to pavement skid resistance. The features extracted from the GIPSICAM georeferenced image data are traffic light controlled intersections, pedestrian crossings, school crossings, railway level crossings and roundabouts. The data is then available for use in safety related projects. 195 Application of MMS Technology 8.3.2 Road pavement crack mapping The RTA has a RoadCrack vehicle, which is capable of accurately measuring the width of cracks in the road pavement and classifying the cracking (Roads and Traffic Authority, 2005 RoadCrack), however no record of crack length is collected and no visual record is maintained. A trial project is about to commence to utilise GIPSICAM for road pavement crack mapping, as a supplemental data source to the RoadCrack data. 8.3.3 Accurate measuring of distance When state roads are created or modified, the network geometry within RAMS is updated to reflect the new state of the road network. A critical attribute required is the length of the new/modified section of road. This length can be obtained by measuring the distance from the start to the end of the new/modified section of road using the georeferenced road images. Another certified “distance measuring” procedure that can be utilised involves using the GCv3 vehicle. The GCv3 vehicle has three independent distance measuring devices, which are a laser-based odometer under the front of the vehicle, a mechanical-based odometer built into the driveshaft, and an odometer fitted to the vehicle ABS system. The section of road or part of a road that needs to be measured very accurately can be driven in the GCv3 vehicle and the three calibrated odometers will provide three sets of accurate, independent and comparative distances. 8.3.4 School bus routes In 2007 and 2008 the NSW Ministry of Transport (MOT) and the RTA undertook a project to survey 25 rural school bus routes. The project had two outcomes, which were the collection of road geometry data and lane width data to be used as part of a MOT 196 Application of MMS Technology school bus route risk assessment, and the collection of road images for planning and as a visual record of the routes. 8.3.5 Road longsections (heights) The alignment data produced by GIPSICAM utilises absolute positions from an OmniSTAR VBS corrected DGPS as anchor points to “tie-in” the relative positioning data acquired from the dead reckoning (DR) system. As mentioned previously, utilising a more accurate source of absolute positions would increase the accuracy of the alignment data that is produced. This concept has been utilised in a recent project for RTA’s Survey Section, using traditional survey methods and GIPSICAM DR data. The project involved the accurate representation of a road centreline longsection, to within 100mm, to be used as a frame-of-reference for a hydrological investigation on a road that was known to have flooding issues. Data was collected at irregular intervals along the road section using traditional survey methods. The survey data was then combined with the GIPSICAM DR data from a GIPSICAM survey of the section of the road to produce an accurate road alignment, from which the longsection information was extracted. 8.3.6 Regional Forest Agreements project National Park Estate legislation was enacted in 2000, 2002 and 2003 which enabled the transfer of title of specific state forest, nature reserve and conservation areas to National Park Estate (Clark, 2005). However, before the transfer of title was to take place, an adjustment of the description of the land to be transferred needed to occur. In particular, the RTA needed to ensure that no classified roads road reserves were included in the Regional Forest Agreement. GIPSICAM centreline data was used as a reference alignment dataset for determining if a road reserve required closer investigation, as well as a verification dataset for data collected by traditional survey methods to accurately determine road reserve boundaries (See Figure 90). 197 Application of MMS Technology Figure 90. Sample GIPSICAM centreline data used in the RFA project 8.3.7 Historical record searches RTA staff regularly source historical GIPSICAM data to determine past road conditions or to verify the presence or absence of a roadside asset. A common search request is “provide all GIPSICAM road images between point A and point B on road C, between the dates of xx/xx/xxxx and yy/yy/yyyy”. Other typical requests include “verify the existence of sign type A at location B on the date xx/xx/xxxx” or “verify the speed limit at location A on the date xx/xx/xxxx”. Historical road geometry data is also requested, with grade and crossfall being the two primary parameters of interest. Use of historical GIPSICAM data by RTA staff is increasing each year, and the GIPSICAM road image and road geometry data has been recognised as an important historical record of the NSW state roads. 198 Application of MMS Technology 8.3.8 SCRIM site categories determination SCRIM is an acronym for “Sideways Force Coefficient Routine Investigation Machine”. SCRIM was developed by the UK Transport Research Laboratory. SCRIM systems are used for monitoring the wet, sideways force skid resistance of sealed road pavement networks (Roads and Traffic Authority, 1996 SCRIM). Skid resistance is an important factor in road safety, thus it needs to be monitored routinely (VicRoads, 2007). However, maintaining a high skid resistance on the entire road network would be expensive, thus each road is broken up into 100m sections and a “potential for an accident” rating, which is referred to as a “SCRIM site category”, is determined for each 100m section, based on the likelihood of an accident occurring in that 100m section due to other factors not related to skid resistance. Then the measured skid resistance is combined with the SCRIM site category for each 100m section and a risk rating is determined. This means that a priority can be given to the 100m sections of road with a high risk rating where an accident is more likely to occur, and ensures that the overall risk of skidding, and ultimately an accident occurring, is reduced across the whole network. A summary of the SCRIM site categories is given below: 50km/h categories: o category 1: . Traffic light controlled intersections . Pedestrian/school crossings . Railway level crossings . Roundabout approaches o category 2: . Curves with radius =< 250m . Gradients => 5% and => 50m long . Freeway/highway on/off ramps o category 3: . Intersections o category 4: 199 Application of MMS Technology . Manoeuvre-free areas of undivided roads o category 5: . Manoeuvre-free areas of divided roads 20km/h categories: o category 6: . Curves with radius =< 100m o category 7: . Roundabouts In the past, SCRIM site categories were “guessed” by the SCRIM operator in the field, a procedure which was highly prone to error. Hence a plan was devised to determine the SCRIM site categories using GIPSICAM data. The grade and horizontal radius information is extracted from the GIPSICAM road geometry data. Intersections, divided/undivided roads and on/off ramps are determined from the RTA road network definitions defined in RAMS. Finally, the positions of traffic light controlled intersections, pedestrian/school crossings, railway level crossings and roundabouts are extracted from the GIPSICAM georeferenced image data. An RTA-developed computer program then automatically determines the SCRIM site categories using the data extracted from GIPSICAM and RAMS. 8.3.9 Point to Point Speed Zones (P2PZ) “Legislation is being introduced in NSW to prosecute drivers that have exceeded the legal speed limit anywhere on specially established road routes, called Point-to-Point Speed Zone (P2PZ)” (Roads and Traffic Authority, 2007 P2PZ). Point-to-point speed camera enforcement is not new, and is currently utilised overseas in places such as the UK. The concept is very simple. Two sets of in-ground detectors and cameras are positioned on a road at a precisely known distance apart. The minimum time to travel from the first detector to the second detector is calculated, with the assumption that the vehicle is driving at the speed limit, using the equation: minimum legal time = minimum distance / speed limit. For example, if the minimum distance 200 Application of MMS Technology between the start and end of the P2PZ is 100km and the speed limit for the entire P2PZ is 100km/h, then the minimum legal time to travel from the start to the end of the P2PZ is 100/100, which is one hour or 60 minutes or 3,600 seconds or 3,600,000 milliseconds. In practice, the way the P2PZ operates is that the start in-ground vehicle detector detects the presence of a motor vehicle as it passes by and records the precise date/time, while a video camera monitoring the detector area extracts and records the licence plate details of the vehicle in question. A known distance down the road the end in-ground detector detects the presence of the same motor vehicle and records the precise date/time, while a video camera verifies the licence plate details. The time taken for the vehicle to travel from the first detection point to the second detection point is then determined, and if this travelled time is less than the pre-calculated minimum legal travel time then the vehicle is sent an infringement notice for speeding. Point-to-point speed cameras were first trialled in NSW in 2004 (SMH, 2004) but the trial was stalled. The RTA’s Camera Enforcement Branch has since recommenced the trial. GIPSICAM is being used to assist the project. Each P2PZ will be surveyed by GIPSICAM to produce a visual record of the P2PZ. The road images will then be used to determine a “measured shortest path” using RTA software. Finally, each P2PZ is surveyed by GIPSICAM a number of times to determine the driven shortest path distance. GIPSICAM has three independent distance measuring devices, which are a laser-based odometer under the front of the vehicle, a mechanical-based odometer built- in to the driveshaft, and finally an odometer fitted to the vehicle ABS system. Once the shortest distance is determined, a buffer margin is added to account for any possible errors. 8.3.10 Speed Limiter Enforcement Zones (SLEZ) “Legislation was amended in November 2005 to allow NSW Police to prosecute those responsible for non-compliant speed limiters, that is, speed limiters which cannot limit heavy vehicle speeds to under 100 km/hr on grades greater than minus 2% (i.e. downhill 201 Application of MMS Technology grade of magnitude less than 2% and all uphill grades). The amendment now places the onus on the person responsible for the vehicle to prove that the speed limiter was in fact compliant. It is possible that that person may in fact challenge, in a court of law, the grade of the road where the vehicle was ‘caught’ by police” (Roads and Traffic Authority, 2006 SLEZ). Initiated by a request from the police, the RTA conducted a project to “determine stretches of major heavy vehicle routes that meet specific criteria of grade, as measured by GIPSITRAC, on which police may confidently ‘catch’ offending vehicles” (Roads and Traffic Authority, 2006 SLEZ). A number of safety margins were built into the project such as a buffer to account for accuracy concerns and the inclusion of a “slow down distance” which takes into consideration other factors such as preceding downhill grades steeper than 2% and tail- winds (Roads and Traffic Authority, 2006 SLEZ). The output from the project was numerous maps showing SLEZ locations and identifying information for determining exactly where each SLEZ starts and ends. 8.4 Concluding remarks Mobile Mapping System (MMS) data is now routinely used throughout the RTA; from the Road Safety Branch using the GIPSICAM data to make roads safer; to the Regional Operations and Engineering Services Directorate using the GIPSICAM data to build, maintain and inspect roads, bridges and other road assets; to the Network Management Directorate using GIPSICAM data to manage traffic and the road network; it can be seen that a MMS such as the RTA’s GIPSICAM is vital technology to a Roads and Traffic Authority. The high number of RTA staff from various directorates who utilise the GIPSICAM data to perform their duties, many on a daily basis, verifies the importance of MMS technology to the RTA. 202 Summary and Recommendations CHAPTER 9 SUMMARY AND RECOMMENDATIONS 9.1 The development of GIPSICAM v3 The aim of the project was to develop the third generation GIPSICAM system, utilising the experience from previous generation GIPSICAM (“Global and Inertial Positioning Systems with Image Capture for Asset Management”) systems, by addressing the objective of designing a Mobile Mapping System (MMS) for the Roads and Traffic Authority NSW, for the purpose of rapid road asset data capture and other applications of MMS technology. This aim has been achieved and is demonstrated by the fact that GCv3 became operational in September 2006 and has been operating successfully for the past five seasons. As a result of this research, the following contributions have been made: A description of the tasks involved in the selection of a vehicle and its modification in preparation for use as a MMS. A description of the tasks involved in the selection of equipment and the integration of subsystems for use within a MMS. A description of the tasks undertaken when conducting calibration procedures, validation procedures, and operational procedures, as required by an operational MMS. Case studies were presented describing the application of MMS technology within a Roads and Traffic Authority. The previous generations of GIPSICAM were discussed and background information on the utilisation of MMS technology by Australian Road Authorities was provided. The criteria for the selection of a vehicle and its modification for use as a MMS platform were discussed. 203 Summary and Recommendations The selection of equipment and the integration of GCv3 subsystems for use within a MMS was investigated. A brief overview of the software utilised by GCv3 was also presented. The calibration procedures, validation procedures, and operational procedures utilised by an operational MMS such as GCv3 were described. Some of the test data collected to verify that the GCv3 system was operating according to required specifications and producing high quality GIPSICAM data were presented. An overview of the application of MMS technology within a Roads and Traffic Authority, including some case studies where the GCv3 system and/or GIPSICAM data were utilised as part of a solution was presented. 9.2 Future research and development opportunities for GIPSICAM Some suggested upgrades to the current GIPSICAM technology are: Utilisation of a calibrated panoramic camera mounted above the camera turret. Calibration of the side camera, and cameras 3 and 4. Replacement of the primary and secondary GPS receivers with receivers that have a multi-GNSS capability (GPS, GLONASS, Galileo, COMPASS, etc). Addition of another fibre optic gyroscope to provide a redundant source of relative positions that does not require daily user calibration. Research into the upgrade of the components and configuration of the GIPSITRAC to current GPS/INS technology. Possible upgrades to configuration include three accelerometers and three gyroscopes, in a 3-axis configuration, with a multi-GNSS capable DGPS, which should increase the reliability and accuracy of the GIPSICAM relative positioning capability and the road geometry sensing capability. Investigation of methods to obviate the need for daily calibration of the GIPSITRAC, including replacement of the GIPSITRAC MEMS sensors for sensors that do not require calibration, such as a fibre optic gyroscope, but is still 204 Summary and Recommendations able to provide road geometry data; the addition of an accurate thermostat control for the rear of the GCv3 vehicle; and the further testing of the actual significance of the variation of the daily calibration values to determine if the daily calibrations are really required or if an average value could be used. Research into the replacement of GIPSITRAC with current GPS/INS technology that does not require calibration and that utilises three accelerometers and three gyroscopes, in a 3-axis configuration, with a multi-GNSS capable DGPS, which should increase the efficiency, reliability and accuracy of the GIPSICAM relative positioning capability and the road geometry sensing capability. Investigation into the use of gyroscope, accelerometer and laser technologies to determine the roll and pitch of the vehicle due to the vehicle’s suspension in real-time during surveys which can then be utilised to offset against the measured road geometry data to determine the actual road geometry data (grade, crossfall and radii). This would remove the need for the static tilt test calibration as the suspension corrections would be calibrated dynamically in real-time and would provide more accurate road geometry data. Utilisation of lasers to profile the road pavement. Utilisation of lasers to determine the horizontal/vertical clearances around the road corridor. Addition of a Ball-Bank meter to determine and record state-wide advisory speeds data. Research into LiDAR technology with a view to incorporating into the GIPSICAM technology. LiDAR technology used in conjunction with video/GPS/INS technology has the potential to provide a more accurate and efficient automated asset inventory capability that far surpasses using video/GPS/INS technology alone. Research into video analytics and image processing to enable the automated and/or semi-automated extraction of asset data during collection in the vehicle or during post-processing in the office. Research into the application of “big data” technologies to MMS technologies to realise the extraction of asset inventory information and analysis of datasets. 205 Summary and Recommendations Upgrade of the GCAM software to provide navigation instructions to the driver and automatically provide real time progress updates back to the office for survey scheduling and for the safety of the vehicle staff. 9.3 Future of MMS technologies within a Roads and Traffic Authority The future for MMS technology within Road and Traffic Authorities is bright. This technology now plays such an important role in modern Road and Traffic Authorities. In particular, once automated asset extraction from road images becomes an operational reality then the focus may shift from the technology to the asset data itself. To be able to rapidly survey a 1000km highway and then have all of the asset and inventory items along that road imported into an asset management system or a geospatial information system within a week will revolutionise the currency and utilisation of asset data. The other consideration, particularly for the NSW RTA is whether it should be developing MMS technology, or outsource MMS development and be just a user of the technology. Advantages of developing the technology include getting delivery of a system that exactly suits the needs of the organisation, as well as having control of the system to ensure the equipment is utilised to best serve the short-term and long-term requirements of the organisation. The disadvantage is that knowledge and resources need to be maintained, which is not the core business of a Road and Traffic Authority. Six years ago when this project started it was the case that no real outsourcing alternative existed, however today there are a number of MMS technologies that have been developed that do offer the possibility of outsourcing the data collection to private companies. The critical factors though revolve around the capability and capacity to provide routine, timely surveys that collect accurate and high quality road image, road centreline and road geometry data, which can be utilised to automatically extract asset data, and populate the data into the organisation’s asset management systems and geospatial information systems. 206 Summary and Recommendations 9.4 Future of MMS technologies within commercial organisations The future for MMS technology within commercial organisations is a realised certainty, with large scale road corridor surveys around the world now undertaken by fleets of MMS vehicles. Since the development of the Aspen Movie Map in 1978, MMS have become more sophisticated and utilise a number of different sensor types providing multiple capabilities to meet a growing list of requirements and applications. Organisations other than Roads and Traffic Authorities have also realised the capabilities and benefits of MMS technologies, which are now applied on a global scale with fleets of MMS vehicles collecting data for all roads across countries and around the world, to be used for applications that include asset management, planning, mapping, advertising, geo -positioning, navigation and social community services. Notable deployments of MMS technology for the rapid, mass collection of road corridor information at a continental or global scale include: Amazon launched “Block View” in January 2005. Block View initially provided panoramic photos of the roads in more than 10 US cities. The coverage increased as more and more cities were surveyed however in September 2006 Amazon made the decision to stop providing the service. What is significant about Block View is that Amazon was the first internet giant to show the power of the MMS capability for advertising and social community services applications. See (Peters, 2005) and (Sterling, 2006). Google developed and released MMS technology called “Street View” in 2007 that initially provided panoramic photos of the roads in several cities in the US. Since 2007 Google has captured Street View panoramic photos in more than 39 countries, including Australia. See (Vincent, 2007) and (Wikipedia, 2011b). Microsoft launched “Streetside” in December 2009, initially providing panoramic photos of the roads in a number of US and Canadian cities. Since 2009 Microsoft has captured Streetside panoramic photos in more than 100 207 Summary and Recommendations cities across the US, Canada and Europe. See (Bing, 2009) and (Microsoft, 2011). ARRB developed the Hawkeye 2000 MMS technologies in the early 2000’s and since then have sold Hawkeye 2000 systems to Roads and Traffic Authorities and commercial organisations around the world. ARRB also conducts annual and ad-hoc surveys for Roads and Traffic Authorities and Local Governments within Australia. See (ARRB, 2010) and (ARRB, 2011). Sensis Pty Ltd procured a small fleet of ARRB Hawkeye 2000 vehicles in the late 2000’s and now use the vehicles to routinely survey the roads in Australia. Sensis do not however make the imagery available to the general public, but instead use the data for asset collection and mapping maintenance purposes to ensure their road corridor spatial datasets are current and up-to-date. They also use the road geometry information as part of their navigation services to commercial clients. Nokia has procured and developed MMS technologies and has a fleet of vehicles around the world, including Australia, surveying the roads and building huge spatial datasets of asset inventories and points of interest. This data is used for the maintenance of their road corridor spatial datasets and navigation services. 208 References REFERENCES ARRB (1995), “GIPSITRAC Roving Station Operational Manual (Draft)”. ARRB (2010), “Hawkeye Systems”, published September 2010. ARRB (2011), “Hawkeye 2000 Series”, https://www.arrb.com.au/Equipment- services/Hawkeye-2000-Series.aspx, Web site last accessed 30 March 2011. Avenir (2004 A), “Avenir HTCH0484 Specifications”, published by Seikou Optical Ltd, http://www.sourcesecurity.com/docs/fullspec/HTCH0484.pdf, Web site last accessed 30 March 2011. Avenir (2004 B), “Avenir HZCH08551 Specifications”, published by Seikou Optical Ltd, http://www.sourcesecurity.com/docs/fullspec/HZCH08551.pdf, Web site last accessed 30 March 2011. Avnet Kopp Pty Ltd (2003), “Piezoelectric Vibrating Gyroscope ENV Series”, Designers Choice, Issue 2, 2003, http://www.avnetkopp.co.za/Designers_Choice/issues/DC2003-02.pdf. Basler Vision Technologies (2005), “Basler A100 series Product Specifications”, published in May 2005 by Basler Vision Technologies, http://www.baslerweb.com/beitraege/unterbeitrag_en_20654.html, Web site last accessed 30 March 2011. Bing (2009), “Bing Maps Adds Streetside, Enhanced Bird’s Eye, Photosynth and More”, published December 2, 2009, https://blogs.bing.com/maps/2009/12/02/bing-maps-adds-streetside-enhanced- birds-eye-photosynth-and-more, Web site last accessed 30 March 2011. Bonsen G, Ammann D, Ammann M, Favey E, Flammant P (2005), “Continuous Navigation Combining GPS with Sensor-Based Dead Reckoning”, GPS World, April 2005, pp. 47-54. Bossler J, Goad C, Johnson P, Novak, K (1991), “GPS and GIS Map the Nation’s Highway”, Geo Info Systems, March 1991, pp. 26-37. 209 References Chee D (2005), “Rural road crash rates database”, IPWEA NSW Division Annual Conference 2005. Clark D (2005), “RFA Process”, GLADS conference 2005. Computar (2006), “Computar H0514-MP Specifications”, published in February 2006 by Computar, http://computarganz.com/file.cfm?id=158, Web site last accessed 30 March 2011. Department of Lands (2005), “LPI Spatial Datasets News”, January 2005, http://www.lpma.nsw.gov.au/__data/assets/pdf_file/0009/25848/LPISpatialDat asetsNews.pdf, Web site last accessed 30 March 2011. Dunlop S (2007), Manager Road Information and Asset Management Technology, Roads and Traffic Authority, personal communication, 4/10/2007. El-Sheimy N (1996), “The Development of VISAT - A Mobile Survey System For GIS Applications”, Thesis Dissertation, The University of Calgary, September 1996. Entriken DR (2005), “GIPSICAM Position Determination Instruments and Methodology”. Internal RTA publication describing in detail the GIPSICAM Position Determination technology, June 2005. Ford T, Hamilton J, Bobye M, Day L (2004), “GPS/MEMS Inertial Integration Methodology and Results”, ION GNSS 2004, 21-24 September 2004. Fujinon (2005 A) “Fujinon DF6HA-1B Specifications”, http://fujinon.com/ProductDocs/DF6HA-1B.pdf, Web site last accessed 30 March 2011. Fujinon (2005 B) “Fujinon HF9HA-1B Specifications”, http://www.fujinon.com/ProductDocs/HF9HA-1B.pdf, Web site last accessed 30 March 2011. Geoscience Australia (2010), “Geodetic Datum of Australia (GDA)”, http://www.ga.gov.au/earth-monitoring/geodesy/geodetic-datums/GDA.html, Web site last accessed 30 March 2011. 210 References Greening S (2003), “GIPSICAM The Capture Management and Application of Terrestrial Road Surveying Imagery in the RTA”. Internal RTA publication summarizing the 2nd generation GIPSICAM technology, 2003. Grejner-Brzezinska D (2004), “Integrated Systems for Mobile Mapping”, presentation notes from a two-day workshop presented at UNSW, 13-14 December 2004. Honeywell International Inc (2004), “Q-Flex® QA-700 Accelerometer Economical temperature-compensated sensor”, http://www.honeywell.com/sites/servlet/com.merx.npoint.servlets.DocumentSe rvlet?docid=DBDA0AD9D-8078-5206-1159-38D9E2550F9D, Web site last accessed 30 March 2011. Kempe V (2011), “Inertial MEMS Principles and Practice”, Cambridge University Press, ISBN 978-0-521-76658-6. Land & Property Management Authority (2006), “Policy for NSW Lambert Conformal Conic Projection”, http://www.lpma.nsw.gov.au/__data/assets/pdf_file/0006/25728/GDA_Lamber t.pdf, Web site last accessed 30 March 2011. Land & Property Management Authority (2010), “Map projections”, http://www.lpma.nsw.gov.au/survey_and_maps/geodesy/projections, Web site last accessed 30 March 2011. Leica Geosystems Inc (1999), “Introduction to GPS (Global Positioning System) v1.0”, Section 3.1.3 Error Sources pp. 14-18. Li Q, Fang Z, Li H (2005), “The Application of Integrated GPS and Dead Reckoning Positioning in Automotive Intelligent Navigation System”, Journal of Global Positioning Systems (2004) Vol. 3, No. 1-2: pp. 183-190. Maerz N, McKenna S (1999), “Mobile Highway Inventory and Measurement System”, Transportation Research Record No. 1690, pp. 135-142. Microsoft (2011), “Explore Streetside”, http://www.microsoft.com/maps/streetside.aspx. Meers P (2007), Road Maintenance Planner, Roads and Traffic Authority, personal communication, 2/10/2007. 211 References Mercedes-Benz (2003) “Mercedes-Benz Sprinter 316 CDI Panel Van specifications”, http://sysdoc.doors.ch/MERCEDES/LCV_Sprinter%20316%20Van%20Spec.p df, Web site last accessed 30 March 2011. Novak K (1993), “Mobile Mapping Systems: New tools for the fast collection of GIS information”, SPIE State-of-the-Art Mapping 1993, 15 October 1993. OmniSTAR Pty Ltd (2011), http://www.omnistar.com.au, Web site last accessed 30 March 2011. Peters K (2005), “Yellow pages go visual with Block View at Amazon’s A9.com”, published January 27, 2005, https://www.internetretailer.com/2005/01/27/yellow-pages-go-visual-with- block-view-at-amazon-s-a9-com, Web site last accessed 30 March 2011. Roads and Traffic Authority (2004 Strategy), “Strategy for Enhancement/Development of RTA Pavement Condition Monitoring Equipment”, Commercial-In- Confidence draft report prepared by TMG International (Australia) Pty Ltd, March 2004. Roads and Traffic Authority (1996 SCRIM), “Guide for the measurement and interpretation of skid resistance using SCRIM”, RTA Technical Procedure TP- GDL-009. Roads and Traffic Authority (2005 RoadCrack), “Determination of extent of road pavement cracking by RoadCrack equipment”, RTA Test Procedure T2160. Roads and Traffic Authority (2006 RGA), “Road Geometry Analyst (RGA) User Guide”. Roads and Traffic Authority (2006 SLEZ), “Heavy Vehicle Speed Limiter Enforcement Zones (SLEZ) Determination using GIPSITRAC”, RTA Quality Procedure. Roads and Traffic Authority (2007 P2PZ), “Point-to-Point Speed Zone (P2PZ) Shortest Possible Path Length (SPPL) Measurements using GIPSICAM”, RTA Technical Procedure TGM-TP-409704. Roads and Traffic Authority (2007), http://www.rta.nsw.gov.au/aboutus/index.html, Web site last accessed 30 March 2011. 212 References Roads and Traffic Authority (2011 GCAP), “GIPSICAM data processing”, technical procedure. Roper P (2003), “Data Collection Technology at ARRB Transport Research”, Road Safety Research, Policing and Education Conference, Sydney, September 2003. Ross B (1983), “Mobile Inventory System”, Alberta Transportation, September 1983, Volume 6, Issue 2, pp. 3-6. Schmeißer F, Dietmayer K (1999), “Rotational Speed Sensors KMI 15/16”, Application Note AN98087, Philips Semiconductors, http://www.nxp.com/documents/application_note/AN98087.pdf, Web site last accessed 30 March 2011. Sony Corporation (2003), “Digital Video Camera Module Technical Manual XCD- SX910CR / X710CR (Color model)”, published in 2003 by Sony Corporation, http://www.alacron.com/clientuploads/directory/Cameras/SONY/xcdsx910_ma nual.pdf, Web site last accessed 30 March 2011. Sony Corporation (2004), “Color Digital Camera Technical Manual DFW-SX910 / X710”, published in 2004 by Sony Corporation, http://www.sony.co.jp/Products/ISP/support/manual/DFW-series_ETM.pdf, Web site last accessed 30 March 2011. Sterling G (2006), “Amazon’s A9 Search Drops Block View & Yellow Pages”, published October 3, 2006, http://www.searchenginejournal.com/amazons-a9- search-drops-block-view-yellow-pages/3858, Web site last accessed 30 March 2011. Sydney Morning Herald (2004), “New speed camera will be watching drivers for 35km”, published February 29, 2004, http://www.smh.com.au/articles/2004/02/28/1077677014366.html?from=storyr hs, Web site last accessed 30 March 2011. Tamron (2004), “Tamron 23FM25SP Specifications”, http://www.tamron.co.jp/en/data/cctv_fa/23fm25sp.html, Web site last accessed 30 March 2011. 213 References Tang J (2007), Road Environment Engineer, Roads and Traffic Authority, personal communication, 5/10/2007. Total Turnkey Solutions (2005), “CMOS and CCD Technology Compared”, http://www.turnkey-solutions.com.au/cam_%20basler_cmos_technology.htm, Web site last accessed 30 March 2011. Trimble Navigation Limited (1997), “Community Base Station System Operating Manual”. Trimble Navigation Limited (2011), http://www.trimble.com, Web site last accessed 30 March 2011. US Naval Observatory (2011), “USNO NAVSTAR Global Positioning System”, http://tycho.usno.navy.mil/gpsinfo.html, Web site last accessed 30 March 2011. VicRoads (2007), “Managing Road Surface Conditions”, VicRoad publication. Vincent L (2007), “Taking Online Maps down to Street Level”, Computer, December 2007, Volume 40, Issue 12, pp. 118-120. Wikipedia (2011a), http://en.wikipedia.org/wiki/Real_Time_Kinematic, Web site last accessed 30 March 2011. Wikipedia (2011b), https://en.wikipedia.org/wiki/Google_Street_View, Web site last accessed 30 March 2011. Williams G (1986), “Alberta Transportation’s Mobile Highway Inventory System”, Alberta Transportation, July 1986. Xiong D, Floyd R (2004), “Highway Feature and Characteristics Database Development Using Commercial Remote Sensing Technologies, Combined with Mobile Mapping, GIS and GPS”. 214 215