A Differential Game Approach for Vehicle Competition and Cooperation at Unsignalized Intersections

This paper studies the navigation strategy of intelligent vehicles crossing through unsignalized intersections. Each vehicle can make decisions independently or act in a cooperative manner. The problem is described as a dynamic differential game in which two vehicles make decisions according to the expected behavior of the other vehicle. Both non-cooperative game (a vehicle only optimizes its own cost) and cooperative game (a vehicle coordinates its decision to optimize a joint cost) are considered. Unlike traditional discrete game theoretical methods that divide the acceleration of vehicles into several pieces, the proposed differential game generates continuous-time acceleration signals that are flexible to vary during the entire intersection crossing period. The efficacy of the differential game approach is verified by simulations, and results show that the method make vehicles pass through the unsignalized intersection more quickly than traditional discrete game theoretical methods, and satisfies safety and comfort requirements at the same time.