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.