A Decision-Making Model for Autonomous Vehicles at Intersections Based on Hierarchical Reinforcement Learning

By aiming at addressing the left-turning problem of an autonomous vehicle considering the oncoming vehicles at an urban unsignallized intersection, a hierarchical reinforcement learning is proposed and a two-layer model is established to study behaviors of left-turning driving. Compared with the conventional decision-making models with a fixed path, the proposed multi-paths decision-making algorithm with horizontal and vertical strategies can improve the efficiency of autonomous vehicles crossing intersections while ensuring safety.