Deep Reinforcement Learning Methods for USV Control: A Review

As an advanced application of the unmanned system in marine engineering, unmanned surface vehicles (USVs) are widely used in the field of sea reconnaissance, mine detection, anti-submarine, water quality monitoring, resource exploration, etc. The control system is the core component of USV, which is a hot topic of research in the field. With the increasing development of artificial intelligence, the use of reinforcement learning has become one of the most popular trends in the decision-making and control of USVs. In this paper, we present a review of reinforcement learning applications in USV's control system in recent years. Firstly, we briefly outline the development history, basic principles, and mainstream algorithms of reinforcement learning. Then, we summarize the research progress of reinforcement learning applied in USV control, including two typical scenarios of motion control and formation control. Finally, future research works are discussed and foreseen according to the problems and challenges of reinforcement learning-based applications.