High Fidelity Localization and Map Building from an Instrumented Probe Vehicle
Total Page:16
File Type:pdf, Size:1020Kb
High Fidelity Localization and Map Building from an Instrumented Probe Vehicle DISSERTATION Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University By Douglas Thornton, B.S., M.S. Graduate Program in Electrical and Computer Engineering The Ohio State University 2017 Dissertation Committee: Dr. Benjamin Coifman, Advisor Dr. Levent Guvenc Dr. Ümit Özgüner Copyrighted by Douglas Thornton 2017 Abstract The lack of high fidelity data sources measuring roadway infrastructure has long handicapped the modeling of vehicular interaction and traffic flow. To date embedded loop detectors and other point detectors provide the data source for these models, aggregating data over significant time intervals, often minutes, performing point measurements at spacing much larger than the characteristics being measured. The result of this inadequacy is a scientific disagreement on even the most basic relationships in traffic flow theory, such as the fundamental relationship between flow, occupancy and density. Beginning in 2005, the Ohio State University began collecting high fidelity traffic flow data from an instrumented probe vehicle. The data mitigates a number of problems of both traditional data sources such as loops, and experimental data sources such as NGSIM, ultimately providing utility to solve or refine a variety of open traffic flow theory problems. As a precursor to applying the instrumented probe vehicle data, raw sensor information must be aggregated and processed using a variety of techniques found in control, transportation, and robotics literature. Data were collected in hundreds of runs over six years under a variety of changes to the environment and sensor suite itself, requiring the data processing to be automated and robust. This research resolves a number of issues with the instrumented probe vehicle data extraction by: 1) providing a method to validate a global localization estimates, 2) ii designing and implementing a new, observational, globally referenced mapping framework and applying that framework to Bayesian occupancy and evidential grid representations, and 3) developing a suite of applications supporting the processing of the data, including LiDAR mounting calibration, localization refinement, map structure change identification, road boundary detection, and lane finding. The novel use of a perception sensor, specifically a vertically scanning LiDAR, solves the issue of verifying a large, historic dataset’s global positioning system derived global localization. This validation supports trust in instrumented probe vehicle by verifying the localization achieves lane level accuracy, as well as in future automated vehicle applications. To aid in the storage and retrieval of observational data of large, city-scale regions, this research creates the Map Oriented Grid, which supports the efficient global referencing of observational data stored in a grid structure. These grid structures support many opportunistic mapping applications including the identification of salient structures nearby a road, and the areas on a roadway where movement if regularly observed. This framework could be applied in crowd sourcing maps in a connected vehicle environment. Finally, a chief goal of the instrumented probe vehicle is to accurately and precisely track the nearby ambient vehicles. The Map Oriented Grid supports such needs by developing applications on top of this framework that were necessary to both simplifying future vehicle trackers based upon this work, and provide the highest quality, calibrated location and sensor data to such a vehicle tracker. In providing the above capabilities, this work assists in the extraction of value from the instrumented probe vehicle data, and correspondingly advances the state of the art in traffic flow theory. iii Acknowledgments I thank my wife Stephanie for all the time and support she has given me while pursuing graduate school, my children Andrew, Maria, and Sarah who have sacrificed time with Dad, my parents David and Bernadette who taught me to love learning, and Dr. Coifman who challenged me to expand my skills in research and writing. iv Vita August 2004 .................................B.S. Electrical Engineering, The Ohio State University June 2009 .....................................M.S. Electrical Engineering, The Ohio State University Publications Coifman, B., M. Wu, K. Redmill, D. Thornton, “Collecting ambient vehicle trajectories from an instrumented probe vehicle: High quality data for microscopic traffic flow studies”, Transportation Research Part C: Emerging Technologies, 72, pp 254-271, 2016. Thornton, D., K. Redmill, B. Coifman, “Automated Parking Surveys from a LIDAR Equipped Vehicle”, Transportation Research Part C: Emerging Technologies, 39, pp 23-35, 2014. Thornton, D., B. Coifman, “Signal Progression Impacts on Transit Buses as Travel Time Probes”, Journal of Transportation Engineering, 141 (8), 2015. Patents 8,257,191 Golf clubs and golf club heads having digital lie and/or other angle measuring equipment 8,500,570 Golf clubs and golf club heads having digital lie and/or other angle measuring equipment 8,801,533 Golf clubs and golf club heads having digital lie and/or other angle measuring equipment v 8,917,051 Integrated fuel processor and fuel cell system control method 8,956,130 Golf clubs and golf club heads 9,403,078 Golf clubs and golf club heads 9,421,429 Golf clubs and golf club heads having digital lie and/or other angle measuring equipment 9,440,127 Golf clubs and golf club heads 9,537,313 Method and system for using demand side resources to provide frequency regulation using a dynamic allocation of energy resources Fields of Study Major Field: Electrical and Computer Engineering vi Table of Contents Abstract ............................................................................................................................. ii Acknowledgements ........................................................................................................... iv Vita ......................................................................................................................................v List of Tables .................................................................................................................... ix List of Figures .....................................................................................................................x Chapter 1: Introduction .......................................................................................................1 1.1. Motivation ................................................................................................................2 1.2. The Instrumented Probe Vehicle .............................................................................4 1.3. Contribution to the IPV Dataset ...............................................................................5 1.4. Contribution to Other Fields ....................................................................................6 1.5. Outline of the Instrumented Probe Van Data Extraction .........................................7 1.6. Summary ..................................................................................................................8 Chapter 2: Global Localization Validation .........................................................................9 2.1. The Need for Independent Validation of Global Localization Results .................. 10 2.2. State of the Practice in Global Localization Validation......................................... 11 2.3. Methodology for Validating Global Localization.................................................. 11 2.4. Application to IPV Localization ............................................................................ 22 2.5. Results .................................................................................................................... 25 2.6. Closing ................................................................................................................... 33 Chapter 3: Map Oriented Grids with Revisiting ............................................................... 36 3.1. The Need for Efficient, Globally Referenced Observational Maps ....................... 38 3.2. State of the Practice in Map Data Representation ................................................. 41 3.3. Development and Application of Map Oriented Grids .......................................... 45 3.4. Results .................................................................................................................... 67 3.5. Closing ................................................................................................................... 77 vii Chapter 4: Applications Utilizing Map Oriented Grids .................................................... 80 4.1. Introduction to the Developed Applications .......................................................... 80 4.2. Perception Sensor Calibration................................................................................ 82 4.3. Localization Refinement ........................................................................................ 97 4.4. Road Boundary Detection .....................................................................................108 4.5. Lane Finding .........................................................................................................110 4.6. Closing ..................................................................................................................114