Accelerating ADAS with Open Source Introduction
• Mikko Hurskainen - ミッコ フルスカイネン • Technical leadership positions in Nokia, Notava, Nomovok, Suunto, Link Motion • Now Technologist in Link Motion. Looking on future technologies. • Mission: making connected cars safe & secure Link Motion
We make CONNECTED CAR COMPUTERS with HARDWARE - OS - APPLICATIONS
Focus on SECURITY 10 years & 200+ with cost efficient Automotive software projects delivered performance
Our HQ is in Finland
2002 2004 2006 2008
2015 2013 2011 2009 What’s ADAS ?
Applications: Applications: • Commuter Applications: • Lane assist • Efficient transport • Manually driven car • Adaptive Cruise control • Local concierge • Collision avoidance
Level 0 Level 1 Level 2 Level 3 Level 4 Level 5 No Driver Partial Conditional High Full Automation Assistance Automation Automation Automation Automation
Applications: Applications: Applications: • Reverse camera • Automated parking • Taxi service • Cruise control • Highway cruiser • Moving office space • Visualization • Platooning • “Ultimate IoT machine”
Classification: SAE Trends & Role of OSS
Enablement of functionality ADAS becoming mainstream with OSS
Level 0 Level 1 Level 2 Level 3 Level 4 Level 5 No Driver Partial Conditional High Full Automation Assistance Automation Automation Automation Automation
Platformisation of ADAS Need for openness systems & cost-efficiency Linux for ADAS – why ?
• Few years ago instrument cluster running Linux was thought not to be possible, now reality • ADAS becoming more complex – deeply embedded designs do not offer structure & re-use that well structured platforms, like Linux, can offer. • Developers prefer desktop platforms. • Linux is evolving into direction that it can be used for safety critical applications. Examples: NXP Linux, OSADL • Linux is POSIX compliant, possible to transfer results to other POSIX platforms. Also hybrid designs possible. Autonomous Driving
Biggest challenge: how to ensure car behaves correctly in ALL situations ? Answer is Open Source: by making algorithms open those can be tested by large community OSS Toolbox for ADAS
Algorithms New components OSS ADAS Platform
OSS SW / communities
Tools, testbench, drivers, platform, Products Automotive Platforms architecture, process, community
Lower cost & faster time-to- market ISO 26262
Safety based view of the system
ASIL Levels
A
B
C
D
ASIL = Severity x (Exposure x Controllability) ASIL = Automotive Safety Integrity Level (ISO 26262) Process Compliancy
Open Source vs ISO 26262 process
• Rigorous reviews • No controlled process • Adherence to good coding • No responsible conventions persons • Well structured • Lack of testing software coverage • External reviews • Lack of design documentation • Known-to-work designs
Need to also take account cyber security (eg SAE J3061) OSS ASIL Maturization
Developers Community Maturization Product development Product
Stable release ASIL pre-qualified ASIL qualified by the release by product by development maturization company community community
Examples: RHEL, OSADL Technologies
Need more: • Sensing • Connectivity • Fusion • ADAS components • Testbenches • Simulation BlueZ • Auditing • Security Evolutionary: OSS Rear Vision Camera
https://github.com/openautocam
Visual guidance
Sensory guidance SAEL0 ASIL A ASIL
Moving object Collision detection recognition
Braking Visualization Collision detection B Drive train L1 control C
Emergency brake L2 Revolutionary: ADAS Processor Connectivity • ADAS processor as brains of a car • Connected to vision, sensors, Camera Cluster & connectivity, Infotainment actuators Radar • Linux as a base ADAS processor instead of embedded
system LIDAR Drivetrain Hybrid Designs
ASIL A / B
ASIL A
Cluster & Infotainment Safety Cluster / Critical Infotainment SW
Microkernel Drive Control ASIL B+
ASIL C / D Less, but more Powerful computers What’s needed ?
Components Community V2X, Sensing, Co-development, Controller information
Process Examples ISO 26262 Reverse camera, compliant cruise control Thank you ! ありがとうございます