Few aspects of modern life inflict as high a cost on society as traffic congestion and automobile accidents. Current work in AI and Intelligent Transportation Systems aims to replace human drivers with autonomous vehicles capable of safely and efficiently navigating through the most hazardous city streets. Once such vehicles are common, interactions between multiple vehicles will be possible. Traffic lights and stop signs, which were designed for human drivers, may no longer be the best method for intersection control. Previously, we made the case for a reservation-based intersection control mechanism designed for autonomous vehicles, but compatible with human drivers. Including human drivers allows incremental deployability as well as support for those who drive for pleasure, but may result in significantly suboptimal performance, as human drivers may be present in dramatically varying proportions. In this paper, we develop a learning-based approach to determine which variant of the control mechanism will be most effective under given conditions, and then combine the resulting predictor with our multiagent intersection management mechanism, enabling it to determine when and how it should alter its configuration to best suit the current traffic conditions. Our extension is fully implemented and tested in simulation, and we provide experimental results demonstrating its efficacy.

PDF, PS, HTML

In AAMAS 2007 Workshop on Adaptive and Learning Agents, pp. 34-39, Honolulu, Hawaii, USA, May 2007.

@InProceedings{2007alag-dresner, title={Learning Policy Selection for Autonomous Intersection Management}, author={Kurt Dresner and Peter Stone}, booktitle={AAMAS 2007 Workshop on Adaptive and Learning Agents}, month={May}, address={Honolulu, Hawaii, USA}, pages={34-39}, url="http://www.cs.utexas.edu/users/ai-lab?2007alag-dresner", year={2007} }