CIS 2011 rkala.99k.org 1 st September, 2011 Planning of Multiple Autonomous Vehicles using RRT Rahul Kala, Kevin Warwick Publication of paper: R. Kala, K. Warwick (2011) Planning of Multiple Autonomous Vehicles using RRT, Proceedings of the 10th IEEE International Conference on Cybernetic Intelligent Systems, Docklands, London, pp
CIS-2011, 1 st September, Autonomous Vehicles Safety Efficient Driving Jam Avoidance CoordinationComfort
CIS-2011, 1 st September, 2011 Software Architecture Sensor Environment understanding Sensor fusion Localization Planning Control Motion MapMission Environment 3
CIS-2011, 1 st September, 2011 Software Architecture 4 Coordination Vehicle 1Vehicle 2Vehicle 3….Vehicle n Inter-Vehicle Communication Coordination Obstacle DiscoveryLocalizationCollision AvoidanceTravel Plan
CIS-2011, 1 st September, 2011 Junior (Stanford) Model 5 Planning
CIS-2011, 1 st September, 2011 Why not speed lanes? 6 Highly diverse sizes Excess road widths with diverse speeds
CIS-2011, 1 st September, 2011 Why not speed lanes? 7 Single lanes Complex obstacle framework
CIS-2011, 1 st September, 2011 Talos (MIT) Model 8 Key Takes: Planner/Controller integrated Possibility of multiple vehicles Source: Y. Kuwata, S. Karaman, J. Teo, E. Frazzoli, J. P. How, G. Fiore, “Real-Time Motion Planning With Applications to Autonomous Urban Driving,” IEEE Trans. Control Syst. Technol., vol.17, no.5, pp , 2009.
CIS-2011, 1 st September, 2011 PLANNING ALGORITHM 9
CIS-2011, 1 st September, 2011 Layers of Planning 10 Route Planning Road Segmentation Road Segment Planning Abstraction
CIS-2011, 1 st September, 2011 Overlapping Segment Planning 11
CIS-2011, 1 st September, 2011 Coordinate Axis System 12
CIS-2011, 1 st September, 2011 RRT 13 Source and direction Goal
CIS-2011, 1 st September, 2011 Curve Generation 14
CIS-2011, 1 st September, 2011 Coordination Principle: Prioritization Rule: Higher priority for vehicle earlier in segment Communication: Trajectory followed stored as hash maps Speed reduced in steps of Δ till feasible plan found 15
CIS-2011, 1 st September, 2011 Results 16
CIS-2011, 1 st September, 2011 RRT Generated 17
CIS-2011, 1 st September, 2011 Vehicle Following 18
CIS-2011, 1 st September, 2011 Overtaking 19
CIS-2011, 1 st September, 2011 Vehicle Avoidance 20
CIS-2011, 1 st September, 2011 Analysis 21
CIS-2011, 1 st September, 2011 Analysis 22
CIS-2011, 1 st September, 2011 Why RRT? * Assuming high resolution maps, traffic scenario, complex obstacle framework. Reduction of resolution not possible 23 PropertyAlgorithm Order* CompletenessGraph > RRT > Evolutionary > Behavioral Ability to generate paths with large number of obstacles Graph > Behavioral > RRT > Evolutionary Path length/ OptimalityEvolutionary > Behavioral > RRT > Graph Computational CostBehavioral < RRT < Evolutionary < Graph
CIS-2011, 1 st September, 2011 Conclusions 24 Problem Planning multiple vehicles Different times of emergence Traffic Scenario Constraints Computational Time Complex Obstacle Framework Solution RRT Road Axis System Curve Smoothening Priority Open Issues Optimality Being un- struck Cooperation Non- autonomous vehicles Uncertainties Curves RRT-Connect
CIS-2011, 1 st September, 2011 THANK YOU Acknowledgements: Commonwealth Scholarship Commission in the United Kingdom British Council 25