Cooperative Trajectory Planning at Unsignalized Intersections Using Deep Reinforcement Learning

Cooperative coordination at unsignalized road in-tersections has attracted increasing interests in recent years. However, most existing investigations either suffer from computational complexity or cannot harness the full potential of the road infrastructure. To this end, we first formulate an optimal coordi-nation problem to maximize the traffic throughput while ensuring the driving safety and long-term stability. To address the key computational challenges and support real-time implementation, we solve this non-convex sequential decision problem by devising a Twin Delayed Deep Deterministic Policy Gradient (TD3)-based strategy in the deep reinforcement learning (DRL) framework. Simulation results show that the proposed strategy could achieve near-optimal performance in sub-static coordination scenarios and significantly improve the traffic throughput in the realistic continuous traffic flow.