强化学习在无人车领域的应用与展望

Unmanned ground vehicle (UGV)can replace human to conduct civilian and military missions, which is of great strategic significance to the construction of intelligent transportation system and the development of army equipment. With the development of artificial intelligence technology, the reinforcement learning technology is regarded as one of the key technologies for UGV to realize intelligent decision making. Firstly, this paper briefly summarizes the development process, basic principles and main algorithms of reinforcement learning. Then, the research progress of reinforcement learning in intelligent decision-making of UGV is analysed and summarized, including obstacle avoidance, lane changing and overtaking, lane keeping, and intersection traffic. Finally, in view of the problems and challenges faced by intelligent decision making based on reinforcement learning, the future research work and potential research directions are discussed and prospected.