Optimal Reciprocal Collision Avoidance ORCA Jur van den
- Slides: 28
Optimal Reciprocal Collision Avoidance (ORCA) Jur van den Berg, Stephen J. Guy, Ming Lin, Dinesh Manocha University of North Carolina at Chapel Hill
Motivation § Robots are becoming cheaper, more mobile, and better sensing § Several mobile robots sharing space is becoming increasingly practical § Our Goal: § Allow robots to share physical space § Encourage smooth, goal directed navigation § Guaranteed collision avoidance http: //gamma. cs. unc. edu/CA 2
Overview § Our Goals § Background & Previous Work § Algorithm Overview § Implementation Details • Performance Results § Conclusions & Future Work http: //gamma. cs. unc. edu/CA 3
Background & Previous Work http: //gamma. cs. unc. edu/CA
Collision Avoidance Static & Dynamic Obstacles § Collision Avoidance is a well studied problem • Velocity Obstacles [Fiorini & Shillier, 98] • Inevitable Collision States[Fraichard & Asama, 98] • Dynamic Window [Fox, Burgard, & Thrun, 97] § Focused on one robot avoiding static and moving obstacles § Inappropriate for “responsive” obstacles http: //gamma. cs. unc. edu/CA 5
Collision Avoidance Responsive Obstacles § Reciprocal Velocity Obstacles(RVO) [Berg et al, ‘ 08] • Extends Velocity Obstacle concept • Oscillation free, guaranteed avoidance (2 agents) § Limitations • Guarantees limited to 2 agents http: //gamma. cs. unc. edu/CA 6
ORCA § A new algorithm for collision avoidance § A linear programming based formulation § Extends Velocity Obstacle concepts • Velocity Based • Provides sufficient conditions for avoiding collisions • Decisions are made independently, w/o communication • Guaranteed avoidance http: //gamma. cs. unc. edu/CA 7
ORCA Algorithmic Details http: //gamma. cs. unc. edu/CA
Problem overview § Inputs: • Independent Robots • Current Velocity of all • Own Desired Velocity (Vpref) § Outputs: • New collision-free velocity (Vout) § Description – Each Robot: • Determines permitted (collision free) velocities • Chooses velocity closest to Vpref which is permitted http: //gamma. cs. unc. edu/CA 9
Velocity Space & Forbidden Regions § Forbidden Regions • Potentially colliding velocities • An “obstacle” in velocity space § VO: Velocity Obstacle [Fiorini & Shiller 98] • Assumes other agent is unresponsive • Appropriate for static & unresponsive obstacles § RVO: Reciprocal VO [van den Berg et al. , 08] • Assumes other agent is mutually cooperating http: //gamma. cs. unc. edu/CA 10
Velocity Obstacle § Time horizon τ § Relative velocities A–B § Relative velocities B–A symmetric in O http: //gamma. cs. unc. edu/CA 11
Permitted Velocities § If velocity of B is v. B • A should choose velocity outside VOA|B {v. B}. § If velocity of B is in set VB • permitted velocities PVA|B(VB) for A are outside VOA|B VB http: //gamma. cs. unc. edu/CA 12
Reciprocally Permitted Velocities § Set VA of velocities for A and set VB of velocities for B are reciprocally permitted if • VA PVA|B(VB) and VB PVB|A(VA) § Set VA of velocities for A and set VB of velocities for B are reciprocally maximal if • VA = PVA|B(VB) and VB = PVB|A(VA) http: //gamma. cs. unc. edu/CA 13
ORCA § u – Vector which escapes VOτA|B • Each robot is responsible for ½u § ORCAτA|B • The set of velocities allowed to A • Sufficient condition for collision avoidance if B chooses from ORCAτA|B http: //gamma. cs. unc. edu/CA 14
Optimality § Infinitely many half plane pairs reciprocally permitted § ORCA chooses plans to: • Maximize velocities “near” current velocities • Fairly distribute permitted velocities between A and B § For any radius r: http: //gamma. cs. unc. edu/CA 15
Multi-Robot Navigation § Choose a velocity inside ALL pair-wise ORCAs § Efficient O(n) implementation w/ Linear Programming http: //gamma. cs. unc. edu/CA 16
Performance Results http: //gamma. cs. unc. edu/CA
Small Scale Simulation (1) § Two robots are asked to swap positions § Generated Path is: • Smooth • Collision free http: //gamma. cs. unc. edu/CA 18
Small Scale Simulation (2) § 5 Robots moving to antipodal points § Smooth, Collision paths result http: //gamma. cs. unc. edu/CA 19
Performance - Scaling § Our performance sales nearly linearly w. r. t. • • Number of Cores Number of Agents http: //gamma. cs. unc. edu/CA 20
Large Scale Simulations § 1, 000 Virtual robots move across a circle § Collision Avoidance is a major component of Crowd Sims. • ORCA can be applied to virtual agents to produce believable motion http: //gamma. cs. unc. edu/CA 21
Conclusion & Future Work § ORCA: • Efficient, decentralized, guaranteed collision avoidance w 3 -5µs per robot • No explicit communication required • Fast running time & smooth, convincing behavior § Future Work • Incorporating kinematic & dynamic constraints • Implement in 3 D environments http: //gamma. cs. unc. edu/CA 22
Acknowledgments § Funding & Support • ARO (Contract W 911 NF-04 -1 -0088) • DARPA/RDECOM (Contracts N 61339 -04 -C-0043 & WR 91 CRB-08 C-0137) • Intel fellowship • Microsoft • National Science Foundation (Award 0636208) http: //gamma. cs. unc. edu/CA 23
Questions? ? http: //gamma. cs. unc. edu/CA 24
Backup Slides http: //gamma. cs. unc. edu/CA
Choosing Vopt § Vopt impacts the robot behavior § Vopt = Vpref • Vpref may not be know • No solution guaranteed to exist § Vopt = 0 • Deadlock likely in dense scenarios § Vopt = Vcur • Nice balance • Vcur ~= Vperf in low density • Vcur ~= 0 in high density http: //gamma. cs. unc. edu/CA 26
Densely Packed Conditions § If Vopt != 0, solution may not exist • Find the “least bad” velocity • Efficient implementation possible with 3 D linear programming http: //gamma. cs. unc. edu/CA 27
Static Obstacles § ORCAs can also be created for obstacles in the environment § ORCA is half-plane tangent to VO τ A|O http: //gamma. cs. unc. edu/CA 28
- Jur van den berg
- Jur van den berg
- Den gode den onde og den grusomme
- Den gode, den onde og den grusomme
- Musik
- Tcas collision avoidance
- Collision avoidance system block diagram
- Gamle karakterskala
- In den kopf schauen
- Ty prid a vladni nam akordy
- Root word jus
- Mestský úrad svätý jur
- [email protected]
- Cum jur
- Jur gerd kirchhoff
- Ref jur
- Brainpop jur
- Describe
- Avoidance risk
- Avoidance risk
- Experiential avoidance
- Outer directed culture
- Deadlock prevention and avoidance
- Churn avoidance
- Types of motivational conflict
- In an avoidance contingency:
- In an avoidance contingency
- Laray m. barna (1997)
- Special emphasis areas