A Game-Theoretic Approach for Minimizing Delays in Autonomous Intersections

Traffic management systems of the near future will be able to exploit communication between vehicles and autonomous traffic control systems to significantly improve the utilization of road networks. In this work, a novel game-theoretic model for the traffic management of vehicles in intersections is introduced. A core concept from game theory that captures the important interplay between independent decision making and centralized control is the notion of a correlated equilibrium. We characterize the correlated equilibria under this model, yielding interesting connections to maximum-weight independent sets in graphs and maximal matchings in bipartite outerplanar graphs. We develop efficient algorithms for computing optimal correlated equilibria and demonstrate through simulations the effectiveness of our algorithms for improving traffic throughput.