Intersection management for autonomous vehicles with vehicle-to-infrastructure communication.

Abstract

This paper proposes an intersection management strategy for autonomous vehicles under the vehicle-to-infrastructure circumstance. All vehicles are supposed to be fully autonomous and can communicate with the intersection management unit to check the traffic situation. Priority of passing the intersection is decided by a static conflict matrix which represents the potential conflict between lanes of different directions and a dynamic information list which could capture the real-time occupation of each lane in the intersection. Compared with the existing approaches in the literature, the intersection management unit in our strategy is more like a database rather than a computational center, and therefore, requires less computational resource and more likely satisfies the real-time requirement in heavy traffic situations. Simulations are conducted using SUMO (Simulation of Urban MObility), in which the proposed strategy is compared with both fixed and adaptive traffic light methods. The results indicate that the proposed strategy could significantly reduce the average time delay caused by the intersection and the corresponding variance, which shows the efficiency and fairness of the proposed strategy in intersection management.