Dipl.-Ing. Felix Lotz. System Architecture & Behavior Planning

Transcription

1 Dipl.-Ing. Felix Lotz System Architecture & Behavior Planning

2 2 System Architecture & Behavior Planning Agenda Motivation and Challenges of Architecture Design PRORETA 3 Functional Architecture Insight into PRORETA3 Behavior Planning Algorithm Summary

3 3 System Architecture Motivation & Challenge System Architecture Hardware Architecture (Functional) Software Architecture Module2 Module1? Module m Module n Bilder:

4 4 System Architecture Motivation & Challenge Challenge 1: Blank sheet of paper Problem with high degrees of freedom Challenge 2: Significance of software architecture Architecture design takes place in early phase of system development Architecture influences the development and implementation efficiency Architecture influences the system understanding and complexity

5 5 System Architecture Motivation & Challenge SLA LDW ACC RTA CVW LKAS Functional Requirements within BAS+ PRORETA3 LCA PAS PDC CMS ICA FSRA PWS TJA ESA BSD TSR EBSA FCW AEB CAV PRORETA 3 Safety Corridor Cooperative Automation

6 based on [Rosenblatt, 1997] 6 System Architecture Motivation & Challenge Direct functional partitioning modules directly depend on sensors and actuators duplication of effort Safety Corridor Cooperative Automation Reactive Architectures Different behaviors compete over vehicle actuators Mode Manager weights votes Arbiter commands Vehicle Controller Command arbitration is required AEB EBSA LDW Hard to optimize in complex environments / situations ESA FCW

7 7 System Architecture Behavior-based Approach for Mobile Robotics Generalized vehicle behaviors Reconfigure the decision process Mode Manager Avoid Obstacle weights votes Arbiter Seek Goal commands Vehicle Controller Avoid Violation Follow Road Change Lane Layered Architectures Hierarchically layered modules, e.g. organizational hierarchy Decomposition of driving task in subtasks Enables deliberate planning to reach long-term goals Avoid Obstacle Behavior Planner Follow Road Seek Goal Change Lane commands Vehicle Controller desired behavior Avoid Violation

8 8 System Architecture Layered Architectures in Praxis Fully automated vehicles [Urmson, 2008] autonomos.inf.fu-berlin.de Unique selling points of PRORETA 3 Intensive driver integration (shared vehicle control) Safety Corridor: Assisted conventional driving Cooperative Automation: Maneuver-based vehicle automation

9 Perception (Sensors) Driver 9 PRORETA3 Architecture Overview (Optional) Mission Layer World Model Planning HMI Environment Representation Topol. Map Environment Interpretation Traffic Metadata Route Plan Route Inf. Mission Planner Destination Route Details Navigation Interface Coordination Layer Behavior Layer Traffic Rule Features Graphbased Map Object List Lane Recognition Occupancy Map Traffic Rule Interpret. Obj.+Feat. Map Object prediction Freespace Inf. Request High Level Inf. Low Level Inf. Guidance Decision Behavior Planner Active Behavior and Status Trajectory Planner Maneuver, Mode Man. Demand Intervention Trajectory Mode Selection Maneuver Interface Driver State System State Feedback Driver Conditioning Trajectory Vehicle Dynamics- and Control Layer Actuator Setpoints [Hohm, Lotz et al. 2014] Actuator Layer + Steering Torque Brake Pressure

10 Perception (Sensors) Driver 10 Example: Safety Corridor Intervention World Model Planning HMI Environment Representation Topol. Map Environment Interpretation Traffic Metadata Mission Planner Navigation Interface Traffic Rule Features Graphbased Map Traffic Rule Interpret. Obj.+Feat. Map Behavior Planner Driver Status Warning Mode Selection Maneuver Interface Driver State Object List Object prediction Intervention Approval Criticality Interv. Recomm. System State Feedback Lane Recognition Occupancy Map Freespace Low Level Inf. Trajectory Planner Driver Conditioning Safe Trajectory Vehicle Dynamics- and Control Layer Actuator Layer Actuator Setpoints + Steering Torque Brake Pressure

11 11 Behavior Planning Exemplary Situation within Cooperative Automation Mode

12 12 Behavior Planning Requirements and Design Questions Mode Change SC KA Approach / Crossing of Intersection: How / when / how long are available maneuvers offered? Is there a default maneuver? What if driver suddenly changes decision? How is yielding represented within implementation? When is the turning maneuver finished?

13 13 Behavior Planner Implementation Requirements: Traffic rules dictate specific vehicle behaviors Driver delegates discrete vehicle maneuvers Implementation approach has to cope with discrete system states Selected Implementation Approach: Hierarchical State Machine Mode Management Mode Selection 2 Determine Possible Modes 1 Safety Corridor Cooperative Automation

14 14 PRORETA 3 System Architecture Summary The modular and layered architecture design allows a division of behavior decision and behavior execution Each layer can be implemented independently Functional encapsulation results in error isolation and reduces the testing and debug effort The driver is explicitly represented within the system design shared vehicle control Optional Mission Layer gives the opportunity to achieve higher degrees of automation without changing the architecture design Planning Mission Planner Behavior Planner Trajectory Planner HMI Navigation Interface Mode Selection Maneuver Interface Driver State System State Feedback Driver Conditioning

15 15 Literaturverweise [Rosenblatt, 1997]: Rosenblatt, J.: DAMN: a distributed architecture for mobile navigation, Journal of Experimental & Theoretical Artificial Intelligence, 2-3, 1997 [Hohm, Lotz et al., 2014]: Hohm, A.; Lotz, F.; Fochler, O.; Lueke, S; Winner, H.:, Automated Driving in Real Traffic: from Current Technical Approaches towards Architectural Perspectives, In: SAE Technical Paper , 2014 [Urmson, 2008]: Urmson, C. et al.: Autonomous Driving in Urban Environments: Boss and the Urban Challenge, in: Journal of Field Robotics, 25(8), 2008 [Geyer, 2013]: Geyer, S.: Entwicklung und Evaluierung eines kooperativen Interaktionskonzepts an Entscheidungspunkten für die teilautomatisierte, manöverbasierte Fahrzeugführung, VDI Fortschrittsberichte, Reihe 12, Nr. 770, Düsseldorf, 2013

16