Komplexitätsreduktion und Beschleunigung: Dr. Michael Reichel 9. November 2016
An ACC Developer s World in 2002. Radar Steering Wheel Angle Motor Speed Control Head Unit Target vehicle candidates, their velocity / acceleration Target vehicle selection Ego vehicle speed control System activation, status communication Items to specify: 4. 2
A Driverless Car Developer s World in 2016. Radar Camera LIDAR Sonar 30 25 20 15 10 5 0 Steering Wheel Sensors Wheel Speeds 1999: Mercedes S-Class Distronic 2002: VW Phaeton ACC IMU / Gyro Global Position Maps Moving objects Items to specify Static obstacles 2006: Audi Q7 ACC plus / AEB Lanes 2005: Mercedes S-Class Distronic plus Signs Vehicle ego motion Trajectory planning & control 2010: Audi A8 GPS-guided ACC Fail-safe / -degraded mechanisms Items to specify: 24. Motor Speed Control 2013: Mercedes S-Class Dist.+ / Steering Assistant 2015: Audi Q7 Traffic Jam Assistant Regenerative Braking Brake Pressure + brakes + sensor fusion + GPS maps + steering + automatic steering 2002 2004 2007 2010 2013 2015 Steering Angle Control Head Unit Side Tasks Functional safety Redundancies 3
# Interactions # Interactions How Many Interactions Can We Expect For n Components? Best case: Every specification gets written or checked only once Worst case: All specifications must be re-examined after one is changed 300 250 200 150 100 50 0 n(n 1) 2 24 components: 276 interactions 1 2 3 4 5 6 7 8 9 101112131415161718192021222324 # Components 9000000 8000000 7000000 6000000 2 n 1 1 24 components: 8.4 million interactions 5000000 4000000 3000000 2000000 1000000 0 1 3 5 7 9 11 13 15 17 19 21 23 # Components 4
So What Does Help? Steering Angle Wheel Speeds Driving IMU / Gyro Maps Camera Global Position Radar Emergency Brake Valet Parking Sonar LIDAR Interactions: n m for n sensors, m functions 5
So What Does Help? Steering Wheel Sensors Vehicle ego motion Moving objects Driving Wheel Speeds IMU/ Gyro Radar Camera Static obstacles Lanes Signs Emergency Brake Valet Parking Global Position Maps LIDAR Sonar Sensor Data Fusion Interactions: n + m + k for n sensors, m functions, k abstraction components 6
From Exponentiality to Linearity. Sensor Data Fusion: From echos to objects and free space HMI Management: From buttons and LEDs to user interactions Driving Emergency Brake Motion Management: From brake pressure to trajectory control Valet Parking Well-defined interfaces Well-defined interfaces 7
Functional Architectures Can we make complicated behavior even simpler? Drive in your own lane! Driving Change lanes if you must! Don t crash! (for heaven s sake!) Come to a safe stop if necessary! 8
Possible Approach: -Based Driving Situative Function Arbitration Arbitrator Applicability Desire Risk Comfort Applicability Desire Risk Comfort Valet Parking Driving in Own Lane HMI Control Trajectory Plan HMI Control Trajectory Plan Coordinator Applicability Desire Risk Comfort Lane Change HMI Control Trajectory Plan Ruleset for Driving Applicability Desire Risk Comfort Emergency Brake HMI Control Trajectory Plan Ruleset for Manual Driving Applicability Desire Risk Comfort Safe State Handling HMI Control Trajectory Plan 10
Architecture Is Key To Managing Complexity. 11
Vehicle Abstraction - Sensors Function Specific Views Vehicle Abstraction - Actuators A Modular Framework for Driving. Sensor Data Fusion Situative Arbitration Motion Management HD Positioning Trajectory Control Object Fusion Grid Fusion Road and Lane Fusion Vehicle Database Longitudinal Control Lateral Control HMI Management Safety and Error Management Interfaces for Environment modeling and Integrated interpretation Extensible safety function concept framework Coordination of vehicle Interfaces control for and HMI interoceptive sensors vehicle localization wheel Introducing ticks, in steering relative system rule-based angle, and health absolute accelerometers coordination monitoring coordinates robinos. and / of gyros functions diagnosis braking, steering, or kinematic full trajectory vehicle control smart environment environment sensors model point clouds, objects, safe-state clear object free upgrade space, triggering lists roads, path from lanes, NCAP intersections, safe to HAD communication by signs adding functions components of vehicle state ADASISv3 for map, interpretation SENSORIS for of typical cloud scenarios, options complex for recognition redundant behavior of emerges environment a-typical from safe scenarios coordinated model hand-over / elementary take-over instrument management behavior cluster and guaranteed functions (e.g. testability minimal / verifyability risk manoeuvers) infotainment display 12
Vehicle Abstraction - Sensors Function Specific Views Vehicle Abstraction - Actuators robinos Key Concepts: Standardized Interfaces Sensor Data Fusion Situative Arbitration Motion Management HD Positioning Trajectory Control Object Fusion Grid Fusion Road/Lane Fusion Vehicle Database Longitudinal Control Lateral Control HMI Management Safety and Error Management Every software component has scalable, documented and standardized interfaces to other components exchangeable interfaces to the base system / OS a pre-industrialized algorithm core 13
Vehicle Abstraction - Sensors Function Specific Views Vehicle Abstraction - Actuators Common best practices accelerate testing Sensor Data Fusion Situative Arbitration Motion Management HD Positioning Trajectory Control Object Fusion Grid Fusion Road/Lane Fusion Vehicle Database Longitudinal Control Lateral Control HMI Management Safety and Error Management
Vehicle Abstraction - Sensors Function Specific Views Vehicle Abstraction - Actuators robinos Key Concepts: Smart Redundancy Sensor Data Fusion HD Positioning Object Fusion Grid Fusion Road/Lane Fusion Vehicle Database configurable sensor mapping Situative Arbitration Grid Fusion Model-free (evidence-based) Dynamic freespace ASIL possible Safety Object and Fusion Error Management Time to collision Time to collision Motion Management Trajectory Control Longitudinal Control Lateral Control HMI Management HMI Management Safety Decision HMI Ctrl Traj. Plan Single point of failure eliminated Model-based (EKF) Classified, dynamic objects ASIL possible 15
A Modular Framework enables you to... Map onto concrete ECU architecture HAD EB modules Your modules 3rd-party modules NCAP Ensure differentiation, shorten time to market Upgrade across models / Upgrade over time 16
Best Practice Leverage: Project, Product Family, Industry Project-Level Architecture Best Practices: Reduce complexity Product Family Architecture Best Practices: Re-use specified, developed, industrialized and tested components Industry-Wide Architecture Best Practices Peer review of functionalities, safety and security mechanisms Lower entry into HAD development No vendor lock-in 17
Open robinos and EB robinos working group members welcome! EB robinos Open robinos implements the open robinos specification provides software modules for prototyping in EB Assist ADTF for rapid embedding on AUTOSAR / DrivePX for production on vehicle ECU developed, tested, verified according to functional safety standards Download the open robinos specification www.open-robinos.com specifies a reference platform for automated driving up to Level 5 (SAE) architecture interfaces data flow control mechanisms software modules functional safety aspects freely available and licensed as Creative Commons Available for download today 18
Q4/2016 More EB robinos components
Thank you! www.eb-robinos.com www.try-eb-robinos.com michael.reichel@elektrobit.com Industry-Wide Architecture Best Practices Peer review of functionalities, safety and security mechanisms Lower entry into HAD development No vendor lock-in 21