“Embedded Vision in ADAS and Autonomous Vehicles: Navigating the New Reality,” a Presentation from Strategy Analytics
0 likes241 views
The document discusses the current landscape and future projections for Advanced Driver Assistance Systems (ADAS) and autonomous vehicles, highlighting the impact of COVID-19, market demands, and technological advancements. It emphasizes the ongoing shifts in automotive architectures and the growing importance of driver monitoring systems while noting that the widespread deployment of Level 4 automation is unlikely until after 2030. Additionally, the analysis reflects on the financial pressures within the automotive sector and suggests a potential rise in joint ventures and partnerships to navigate these challenges.
“Embedded Vision in ADAS and Autonomous Vehicles: Navigating the New Reality,” a Presentation from Strategy Analytics
- 1. © 2020 Strategy Analytics The Road from ADAS to Autonomous Vehicles: Navigating the New Reality Mark Fitzgerald Director, Autonomous Vehicles
- 2. © 2020 Strategy Analytics Agenda • One Page Strategy Analytics Overview • What is happening in ADAS and Autonomous Driving? • Sensors: ADAS and Autonomous Driving Trends • Automotive Architectures • Q&A 2
- 3. © 2020 Strategy Analytics Strategy Analytics Overview 3
- 4. © 2020 Strategy Analytics Strategy Analytics Overview • 12-month syndicated subscription services on specific industries and/or technologies • Custom Consulting on Companies, Brands, Products, and Technologies • Market Intelligence on Buyer Behaviors, Consumer Attitudes, Brand Preferences, and Emerging Behaviors • Design Guidance and Competitive Intelligence on User Experiences and Opportunities for Innovation • Intelligence on Consumer Activities, Behavioral Patterns, and Usage Profiles through Big Data Analytics 4
- 5. © 2020 Strategy Analytics What is Happening in ADAS and Autonomous Driving? 5
- 6. © 2020 Strategy Analytics AV ecosystem Strong signs of “reality” breaking out at many companies; COVID-19 not helping… Expect many more JVs / mergers / acquisitions and re- focusings – especially when it comes to monetizing AV development for ADAS • E.g. Volvo / Zenuity / Veoneer A falling car market and static ADAS market will put severe pressure on investment levels in 2020 and beyond Big infrastructure investment (e.g. V2X) also now potentially even more problematic 6
- 7. © 2020 Strategy Analytics What about L2+ ? So-called L2+ now seen as a path-to-market by many: • Optimist’s view: Let’s make cars as safe and automated as possible under current legal frameworks • Cynic’s view: We’ve spent all this money, how on earth are we going to get a return this side of 2030? Remember: • L2+ and now L2++ have no formal definition – marketed as a consumer convenience – consumers are not familiar with SAE levels of autonomy • In many cases it appears to be L3/L4-style technology (and hence cost), but with the driver still required to supervise on a continuous basis • Audi abandoned L3 Traffic Jam Pilot on its current A8 model. • How many consumers will pay for this functionality? Subscription • Older drivers are the least enthusiastic 7
- 8. © 2020 Strategy Analytics Global adas demand $54B by 2026 • Best combination of growth/size remains in Distance Warning • AEB is the key feature in that category • V2X forecast reduced in latest update • AFL = Adaptive Front Lighting Source: AVS Strategy Analytics Apr 2020 8
- 9. © 2020 Strategy Analytics What is a Realistic autonomous deployment scenario ? Source: AVS Strategy Analytics Nov 2019 Standard Scenario 9
- 10. © 2020 Strategy Analytics Why is this important? What’s at stake? Strategy Analytics expects very little volume of L4 until 2030+ • Our recent update has seen a delay to the uptake of L4 • Most research contacts now more pessimistic L2 to expected to start to emerge in high volume soon • The lack of a formal definition of L2+ makes it too nebulous to forecast with accuracy at present • However, it has the theoretical potential to run to tens of millions of vehicles in 2030, and thus billions of dollars of revenue for suppliers and automakers Many automakers see adding automation options to their vehicles as a vital way of recouping the very real cost of adding on increasingly-mandated ADAS features • Getting these automation features wrong will thus have a measurable impact on profitability Source: AVS Strategy Analytics Nov 2019 10
- 11. © 2020 Strategy Analytics Sensors: ADAS and Autonomous Driving Trends 11
- 12. © 2020 Strategy Analytics Covid-19 impact on auto cameras Standard view has all the gains from increased camera penetration rates in 2020 wiped out by production falls • Over 100M “lost” cameras from 2020 to 2026 Pessimistic view sees auto camera volume demand fall by 5% over 2019 • Over 200M “lost” cameras from 2020 to 2026 12
- 13. © 2020 Strategy Analytics Euro NCAP Roadmap 2025 Country/Region NCAP/Mandate System 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 European Union NCAP AEBS/Front Corner A. 5* Pedestrian Bicycle Crossing, Junction & Head-On European Union NCAP LDWS/Evasive Steer. 5* Lane Kp.Road Edge Evasive Steering European Union NCAP BSD European Union NCAP Speed Alert-TSR European Union NCAP Backup Camera AEBS Reversing European Union NCAP DMS 5* European Union NCAP Rear PCS/Whiplash European Union Mandate AEBS New models All new vehicles European Union Mandate Alcohol Interlock* New models All new vehicles European Union Mandate Backup Camera New models All new vehicles European Union Mandate DMS New models All new vehicles European Union Mandate Event Data Recorder New models All new vehicles European Union Mandate LKAS New models All new vehicles European Union Mandate Speed Assist New models All new vehicles European Union NCAP Child Presence Det. European Union NCAP Pedestrian Protect. Enhanced European Union NCAP V2X • Initial DMS are systems that infer driver capability from movements of the host vehicle will be sufficient. Such inferred systems sense vehicle movement from existing on-board sensors, such as accelerometers, steering angle sensors and front windshield cameras. • However, members of the UNECE safety committee believe that, by 2022, the test protocols from Euro-NCAP will be tightened to include direct monitoring of the driver’s eyes and face movements – and thus could be beneficial for interior camera-based DMS. ’Driver fatigue and distraction can be major factors in accident causation and can be detected directly by eye-monitoring sensors, for example, or indirectly by identifying driving behaviors which are characteristic of an impaired driver.’ 13
- 14. © 2020 Strategy Analytics Legislation & Mandates Drive ADAS NCAP/Mandate System 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 NHTSA/IIHS AEB 5* Recommended NCAP FCW 5* Recommended NCAP LDWS 5* Recommended Mandate Backup Camera New models All new vehicles Mandate Proposal Child Presence Det. Mandate Proposal FCW NHTSA Agreement AEB NHTSA Agreement Child Presence Det. March 2020 NTSB Recommendations: • For vehicles equipped with Level 2 automation, work with SAE International to develop performance standards for driver monitoring systems that will minimize driver disengagement, prevent automation complacency, and account for foreseeable misuse of the automation. • After developing the performance standards for driver monitoring systems recommended in Safety Recommendation H- 20-X, require that all new passenger vehicles with Level 2 automation be equipped with a driver monitoring system that meets these standards. 14
- 15. © 2020 Strategy Analytics Driver monitoring Strategy Analytics sees the market following three generations: 1. Solutions on dedicated hardware – this is where we are now 2. Solutions that are effectively “software only”, and which are hosted on a shared ECU, e.g. ADAS domain controller or cockpit domain controller 3. Transition from DMS to multi-seat occupant detection system (ODS), especially for AV / robo-taxi applications Conventional camera (2D monocular cameras, used in conjunction with NIR LEDs) remains the preferred approach. ToF-based sensors remain niche for now (costly, bulky, lack of resolution vs. camera based solutions), as will RADAR and other solutions. For Gen 1 & Gen 2 the KEY task is driver monitoring • Everything else (ID, emotion, health) is secondary. “Every OEM is asking for emotion analysis, but no-one seems to know what they want to do with it” – EU-based algorithm vendor 15
- 16. © 2020 Strategy Analytics Market development of optical DMS Camera-based solutions forecast to deploy rapidly starting now Legislative / NCAP interest growing… Forecast has thus been significantly boosted over recent updates Post-COVID-19 16
- 17. © 2020 Strategy Analytics Automotive Architectures 17
- 18. © 2020 Strategy Analytics Key change areas for vehicle architectures Central ADAS Domain Controllers Cockpit Domain Controllers Gateway Modules (A stand-alone module that links vehicle bus networks, e.g. low/hi-speed CAN, FlexRay etc.), effectively acting as a router and firewall for the vehicle. “Full” centralization still appears very much “next-gen” or beyond for mass-market vehicles. Next-stage evolution will most likely be to location-based controllers •Reduce ~ 70 ECUs/modules computers to three. •Each powerful enough to handle the processing necessary to run its portion of the automobile •Software largely developed in- house - has proven to be difficult •OTA updates •Single electronic architecture 2019 Big News - Volkswagen's goal: Source: NXP 18
- 19. © 2020 Strategy Analytics Current vehicle architectures 19
- 20. © 2020 Strategy Analytics The Impact of Vehicle Architectures Stage #1: domain controllers Centralized architecture allows automakers to leverage advancements in processing, memory, software and high bandwidth networking technologies to meet future ADAS and autonomous driving performance requirements such as: reliability/fault tolerance, latency, redundancy, security, flexibility/updatability/OTA updates and cost. 20
- 21. © 2020 Strategy Analytics Location-based architecture • Control centralized based on the location in the vehicle of the function, reducing cabling costs and weight • Much more cross-partner integration required 21
- 22. © 2020 Strategy Analytics To centralize or not? Status quo • Understood technology • Lower risk, despite complexity • Allows extra hardware cost to only be carried when feature is present • “Easy” / “Legacy” relationships with suppliers • Less need for OEM to act as integrator Centralize • Cost-saving potential if most features to be centralized are standard and there is little “wasted” CPU/Memory • Performance benefits – working with “raw” data • More flexibility on sourcing functions as they increasingly become “software only” • New OEMs have no legacy to worry about 22
- 23. © 2020 Strategy Analytics Cockpit EE architecture trend and timeline Distributed EE Architecture (limited domain consolidation, primarily via CDCs) Domain Controller Architecture (number of CDCs increases, despite large number of headunits in market) Location or zone-based EE Architecture (select, premium OEMs begin shift to zone-based architecture) Fully Centralized Processing Architecture (future proposed architecture; unknown if auto industry will actually embrace this approach) 2020 - 2022 2023 – 2027 2027 – 203X 20XX? 23
- 24. © 2020 Strategy Analytics Conclusions Full impact of COVID-19 still far from certain– PLANNING FOR OVERALL VEHICLE SALES/PRODUCTION TO BE IN THE ORDER OF 15% BELOW 2019 LEVELS SEEMS PRUDENT ADAS will be hit less then many other application areas, as it is still a growth market in penetration terms – TOTAL ADAS DEMAND IN 2020 LIKELY TO BE VERY SIMILAR TO 2019 For perception-related technologies, where will the revenue be in the next 5 years? DMS and front facing cameras will see growth COVID-19 is accelerating an already present industry trend that is seeing power shift away from traditional automotive players – WHO SHOULD YOU PARTNER WITH? AV timelines were already receding pre-COVID-19, and the crisis is only pushing launch dates further out (for most) – BIG POTENTIAL WINS FOR THE BRAVE AND WELL-FUNDED The key decision points for NOW are: DO I JUST NEED TO SURVIVE ? …or… CAN I TAKE ADVANTAGE OF OTHERS’ WEAKNESS? 24
- 25. © 2020 Strategy Analytics Questions and Answers 25
- 26. © 2020 Strategy Analytics Resource Slide 26 Freely-available Information Blogs https://www.strategyanalytics.com/strategy- analytics/blogs/automotive/autonomous-vehicles https://www.strategyanalytics.com/strategy- analytics/blogs/automotive/in-vehicle-ux Complimentary Research https://www.strategyanalytics.com/access- services/automotive/autonomous-vehicles/complimentary https://www.strategyanalytics.com/access- services/automotive/in-vehicle-ux/complimentary Paid-for Content ADAS/Autonomous Forecasts & Databases https://www.strategyanalytics.com/access- services/automotive/autonomous-vehicles/market-data ADAS/Autonomous Written Analysis https://www.strategyanalytics.com/access- services/automotive/autonomous-vehicles/reports In-Vehicle UX Reports & Analysis https://www.strategyanalytics.com/access- services/automotive/in-vehicle-ux/reports