Motion Planning of Vehicle Obstacle Avoidance by Quantum Reinforcement Learning

With the development of artificial intelligence, autonomous driving technology is becoming a promising solution for reducing the number of accidents caused by human factors of drivers. Motion planning of vehicle obstacle avoidance is a key part of autonomous driving technology, re-planning the driving path or trajectory based on real-time traffic information. How to ensure the safety of autonomous vehicles (AVs) within limited computing resource is one of the research difficulties. In this paper, a vehicle motion planning algorithm in the hierarchical quantum reinforcement learning (HQRL) framework that combines quantum reinforcement learning (QRL) and hierarchical reinforcement learning (HRL) is proposed. In the option-critic framework of HRL, the QRL agent is applied to make choice in different options. The experimental results showcase our algorithm has better performance on safety average scores compared to classical HRL in the case of less number of model parameters. Therefore, this paper shows the potential of HQRL on solving vehicle motion planing safety problems.