1. Steering Behaviors for Autonomous Vehicles in Virtual Evironments Hongling Wang Joseph K. Kearney James Cremer Department Of Computer Science University of Iowa Peter Willemsen School of Computing University of Utah

2. Focus Control of Autonomous Vehicles in VE –Ambient traffic –Principal roles in scenarios Importance of Road Representation –Frame of reference –Natural coordinate system Intersection and Lane Changing Behaviors –Complex interactions among vehicles Limits of independent control

3. Motivation VE as Laboratories for Studying Human Behavior –Developmental differences in road crossing –The influence of disease, drugs, and disabilities –Design of in-vehicle technology Cell phones, navigation aids, collision warning

4. Bicycle Simulator Video

5. Gap Acceptance in the Hank Bicycle Simulator

6. Related Work Flocking –Complex group behavior from simple rule-based behaviors (Reynolds) Hierarchical Distributed Contol –Independent, goal-oriented sub-behaviors (Badler et al.; Blumberg and Galyean; Cremer, Kearney, and Papelis) Driving –Simulation (Donikian; Lemessi) –ALV (Coulter, Sukthankar; Wit, Crane, and Armstrong) –Human Driving Behavior (Ahmed; Boer, Kuge, and Yamamura; Fang, Pham, and Kobayashi; Salvucci and Liu)

7. Roadway Modeling Roads as Ribbons –Oriented Surface –Smooth Strips –Twist and turn in space Central Axis –Arc-length parameterized curve Twist Angle Linked through Intersections

8. Ribbon Ribbon coordinate system –Distance, Offset, and loft (D,O,L) Egocentric frame of reference Efficient Mapping (D,O,L) (X,Y,Z)

9. Intersections—Where Roads Join Shared regions Non-oriented Corridors connect incoming and outgoing lanes –Single lane ribbons –Annotated with right-of-way rules

10. Ribbon to Ribbon Transitions Problem: Tangle of Ribbons Bookkeeping Tedious and Error Prone Possible switch in orientation Possible shift in alignment Solution: Paths Composite ribbons

11. Path One-lane Overlay –Removes transitions between ribbons Immediate Plan of Action -Highly dynamic -Natural frame of reference

12. Distributed Control Multiple, Independent Controllers –Each responsible for some aspect of behavior e.g. Cruising, Following –Compete for control Control Parameters –Acceleration –Steering Angle

13. Road Tracking Non-holonomic constraint Rolling wheels Move on a circle Pursuit point control –Steer to a point on the path –Look-ahead distance

14. Controlling Speed Cruising: Proportion Control Following: Proportional Derivative Controller

15. Intersection Behavior Gates access to shared regions – Decision: Go / No Go – Action: stop at stopline

16. Gap Acceptance Based on Interval Analysis –Right-of-way rules encoded in DB –Corridors as resources Compare crossing intervals time c0c0 t enter t exit c2c2 c1c1

17. Intersection Exceptions Problem: deadlock Double blocked threats Solution: Recognition and response Problem: starvation Unending stream of opposition Solution: Guaranteed progress

18. What’s missing? Where do paths come from? –Vehicles meander Pick corridors Add outgoing road No goal seeking behavior –Need directions “Turn right at the first intersection, drive through two intersections, and then turn left.”

19. Route A succession of roads and intersections –Like MapQuest Directions A global, strategic goal –The path must conform to the route May require lane changes

20. Stages of Lane Changing Motivation Why change lanes? Decision Choosing a target lane Deciding when to go Action How to change lanes?

21. Motivation to Change Lanes –Discretionary Lane Change (DLC) to improve driving conditions (e.g. speed, density) –Mandatory Lane Change (MLC) to meet destination requirements (e.g. lane termination) Decision to Initiate a Lane Change –Best conditions (e.g. flow) –Gap Acceptance Lead gap Lag gap

22. Lane Changing Action Shift Pursuit Point –Proportional Derivative Controller –Speed Coupling

23. Behavior Combination Combine accelerations from –Cruising behavior –Following behavior –Intersection behavior Combine steering angle from –Tracking behavior –Lane changing behavior

24. Interactions Between Controllers Problem: impeded progress Following prevents overtaking Solution: Reduce following distance Stiffen controller Problem: unveiled threat Appearance of leader in new lane Solution: Split attention – follow 2 leaders

25. Summary An accurate, efficient, robust roadway model –Ribbon network –Arc length parameterization –Efficient mapping between ribbon and Cartesian coordinates A framework for modeling behaviors –Ribbon based tracking –Path based behaviors –Route as a strategic goal

26. Future Work Pedestrians Modeling non-oriented navigable surfaces (e.g. intersections) Pursuit Point Control Behavioral Diversity

27. Acknowledgments NSF Support: INT-9724746, EIA-0130864, and IIS-0002535 Contributing students, staff, faculty Jodie PlumertGeb Thomas David SchwebelPete Willemsen Penney Nichols-WhiteheadHongLing Wang Jennifer LeeSteffan Munteanu Sarah RainsJoan Severson Sara KoschmederTom Drewes Ben FragaForrest Meggers Kim SchroederPaul Debbins Stephanie DawesBohong Zhang Lloyd FreiZhi-hong Wang Keith Miller Xiao-Qian Jiang