from adas to the fully autonomous vehicle: what is the
TRANSCRIPT
From ADAS to the fully autonomous vehicle: What is the reality?
Alain Dunoyer – Head of Safe Car Division
• Driver Monitoring system introduced on the Lexus
• Volvo introduced Collision warning with Auto Brake and Blind Spot Monitoring
• Audi began offering Lane Assist system
• 30% of models offered with ADAS either as standard or optional fitments
• In late 2012 Fiat, Seat an Skoda introduced AEB City system
• Mercedes- Benz introduced Pre-Safe Brake collision avoidance system
• Citroen introduced infrared sensor based LDWS
Safe Car Market – Historical View (EU)
• BMW introduced ACC
• Less than 5% of models offered with ADAS
• Mercedes- Benz and Jaguar started offering ACC
Advanced Driving Assistance - Landscape
Visualisation Information Activation
ADAS Scene Viewing (ADAS-SV)
ADAS Intelligent Driving Support (ADAS-IDS)
Parking Aid
Junction View
Birds-eye View
Night Vision (Gen1)
Night Vision (Gen2)
Night Vision (Gen3)
Driver Monitoring
Lane Departure Warning
Forward Collision Warning
Traffic Sign Recognition
Blind Spot Monitoring
Cross Traffic Alert
Lane Change Merge Aid
Lane Keeping Assist
Automatic Emergency
Braking
Adaptive Cruise Control
Blind Spot Intervention
Cross Traffic Intervention
Automatic Headlamp
Dipping
Beam Shaping
Advanced Driving Assistance - Trends
Visualisation Information Activation
ADAS Scene Viewing (ADAS-SV)
ADAS Intelligent Driving Support (ADAS-IDS)
Parking Aid Parking Aid
Junction View
Birds-eye View
Night Vision (Gen1)
Night Vision (Gen2)
Night Vision (Gen3)
Driver Monitoring
Lane Departure Warning
Forward Collision Warning
Traffic Sign Recognition
Blind Spot Monitoring
Cross Traffic Alert
Lane Change Merge Aid
Lane Keeping Assist
Automatic Emergency
Braking
Adaptive Cruise Control
Blind Spot Intervention
Cross Traffic Intervention
Automatic Headlamp
Dipping
Beam Shaping
Parking Aid
Trend 1 Move
towards automation
Level 3
Lane Departure Warning
Forward Collision Warning
Traffic Sign Recognition
Lane Keeping Assist
Automatic Emergency
Braking
Adaptive Cruise Control
Trend 2 Move
towards automation
Level 3
Blind Spot Monitoring
Cross Traffic Alert
Lane Change Merge Aid
Blind Spot Intervention
Cross Traffic Intervention
Trend 3 Lateral warning has shifted to
radar technology
Night Vision (Gen3)
Automatic Headlamp
Dipping
Beam Shaping
Night Vision (Gen1)
Night Vision (Gen2) Vs
Trend 4 Competing night vision solutions
Automotive ADAS Sensor Maze
CHARACTERISTICS
FOV, Min & Max Range
Fixed Beam/Scanning
FOV, Min & Max Range
Imager type (CCD, CMOS, IR (Far and Near), HDR etc.)
Bandwidth
Ultra Wide Band
Fixed Beam / Scanned
FOV, Min & Max Range
B&W, RBC, RBG, etc.
FOV, imager size, pixel size etc.
Ultrasonic
Cameras
Radar
Laser
Map
V2X
SENSOR TYPE
Mono Vision
Stereovision
24 GHz
76 GHz
DSRC
LTE
TECHNOLOGY TYPE
Sensor selection has a
very significant impact on
feature performance
& end-use experience.
For an OEM, choosing the
“right sensor(s)” is
far from straight
forward. This is very much a balancing act.
What do Sensors Need to Identify?
Target Parameters
Types of sensors
77 GHz 24 GHz Mono
camera Stereo camera
Long range Lidar
Short range Lidar
Ultrasonic
Where is it? Good Average Average Average Good Average Poor to Good*
What is it? Poor Poor Good Good Average Poor Poor
How big is it? Poor Poor Average Good to
Average* Good Poor Poor
How close is it ? Good Good Average Average Good Good Poor to Good*
How fast am I approaching the object ?
Good Average Average Average Average Average Poor to
Average*
* Depending on the distance and /or the application
How fast am I approaching it?
Where is it ?
What is it ?
How big is it ?
How close is it ?
Sensor Selection: a Balancing Act
0
1
2
3Piece Price cost
Scalability
Packaging
ServiceComplexity
Range Accuracy
Range RateAccuracy
Angular Accuracy
ObjectClassification
Bad LightingPerformance
Bad WeatherPerformance
Cost of Ownership
77 Ghz Radar
Monovision (Visible range)
Long Range Laser
ADAS - OEMs Sensor Strategies
Lowcost radar on theirentry level models qualifyingEuroNCAP new AEB test rating;almost 47% reduction in thefeature price from €555 in2012 to €295 in2013.
Low cost radarbased AEB on entry level;
radar & camera fusion unit onpremium models.
Offering low cost singlecamera solution for most
ADAS on entry level models;more expensive radar &camera solution on their
premium models.
In 2012 the E-Class wasoffered with radar basedAEB at €2678, while in 2013the model is offered withradar + stereo camera fusion
unit for the same price.
E x p e n s i v eradar & camera ontheir entry level models;very expensive radar & stereocamera fusion unit onpremium vehicles.
S t i l lthe only
OEM to offer laserbased AEB as standard
on over 90% of its models; moreexpensive radar + camera fusion unit
offered on most models withsmall price increase of
€100 since 2012.
EuroNCAP new testing regime has fuelled the deployment of ADAS, especially in
the collision avoidance area. The solutions deployed are varied and very much
depending on the OEM segment, brand value and longer term strategy.
The Rise of the Camera Systems
Cameras for Scene Viewing
• Such applications can only use camera solutions and are being deployed very rapidly
• Reversing cameras are becoming standardized (Legal push in the USA)
• 360 views are now spreading to the lower segments
• Added value via object detection is appearing
Cameras for Intelligent Driving Support
• Radar sensors dominate both front (EuroNCAP) and rear facing applications, however
camera are being added to the mix, either to increase feature level and/or to improve
robustness.
• Nearly 10 different combinations of sensors (both single and fusion) options are being
used for implementing AEB.
• Infrared camera sensing technology is still very much limited to Night Vision systems
and is competing with advanced lighting technologies…
• https://www.youtube.com/watch?v=Tajs8sqfOQY
Camera System: Existing Challenges
Cameras for Scene Viewing
• Contamination (dust, water droplets etc.)
• Night time performance (blooming, noise etc…)
• Image quality (color aberrations, contrast, resolution, etc.)
• Becoming a commodity from an OEM perspective (CES influence)
Cameras for Intelligent Driving Support
• Limited range (depending on object to be tracked) compared to Radar sensors
• Night time performance
• Bad weather performance
• OEM Business case (feature bundle value – ACC/LDW/TSR/AEB)
Camera Systems: What the Future Holds
Cameras for Scene Viewing
• More advanced visualization (better stitching, less distortion,
innovative viewing angles etc.)
• Computer vision / image processing capability to deliver
new features will become a differentiator
Cameras for Intelligent Driving Support
• Object detection capability in bad weather & bad lighting conditions will be needed
(Future EuroNCAP requirements)
• An effective driver monitoring is needed!
• Identifying “free path” will enable higher level of automation
• Supporting higher levels of SIL & complying with ISO 26262
Scene viewing and intelligent driving support features have been developed
under very different conditions (convenience vs safety). With the introduction of
new electrical architectures, some OEMs will be looking into merging both
systems within their Infotainment platform or within a dedicated ADAS ECU.
Automated Driving Tasks Today
“Dynamic” scenario
Brief event
Highly dynamic
Dynamic Stability Control
Panic Brake Assist
Collision Mitigation & Avoidance
Self-Parking
“Cruising” scenario
Prolonged event
Slow dynamic - steady state
Cruise Control
Adaptive Cruise control
Lane Keeping
Traffic Jam Assist
The machine takes over when the human has
failed to react or is failing to manoeuvre.
The machine helps / support in tedious,
repetitive tasks demanding constant
attention
Automating the Driving Task Further
Automation can lead to monotony (Cruise
control study by Vinci Autoroutes – July 2013)
and in turn increase drivers reaction time…
- Coping with the unknown -
Not being able to regain control
Roadwork
Unexpected scenarios
Debris
Etc.
Automation on a large scale, what is needed?
• Misuse
• Unexpected events
• Fail Safe (Mechanical / Electrical / Software)
• Legality
• Insurance Industry
• Liability
• Affordable
• More capable (all weather, longer range, etc.)
• Packageable
• Is there one?
• Societal benefits are not clear
• Can a human supervise an autonomous vehicle?
• Human Machine Interface
• Hand-over procedure
Robustness
Legal Framework
Sensors
Business Case
Human Factors
ADAS Challenges & Opportunities
• Robustness & Reliability
• Misuse
• Security
• Cost benefit
• More customer demand (Comfort features)
• Increased Safety
• Reduced congestion
• Reduced pollution
• In the near future increased level of automation for limited scenarios are going to get deployed (e.g. parking).
• Customer acceptance & usage is yet to be understood and will be key to further level of automation
Thank you for listening.
Any Questions?
Alain Dunoyer
+44 (0)1908 305107
Free ADAS news service
http://safecarnews.com