An Efficient Collaborative Framework For Connected and Automated Vehicles at Unsignalized Intersections under Various Traffic Loads

Abstract

The research on V2X-related technologies provides new ideas and challenges for solving collaborative vehicle optimization and right-of-way management at intersections. For the purpose of maximizing traffic throughput at the intersection under various traffic loads, a cooperative grouping strategy combined with the Monte Carlo tree search(MCTS) is proposed, which finds the approximate optimal passing sequence by grouping vehicles that can pass through the conflicting area simultaneously. We exploit several numerical simulation experiments to investigate the effect of traffic flow and the number of lanes on traffic efficiency. Simulation results show that the efficiency improvement of this new strategy can reach a maximum of 70% in high traffic load scenarios compared with the other three methods within 100ms planning time.