关键词: 无人艇；编队组织；动态避障；集群；模型预测；强化学习

Abstract: At present, marine vehicles are showing development trends such as intelligence, networking, and clustering. Obstacle avoidance technology for formation in complex environments has currently become the focus, and dynamic obstacle avoidance technology is the key difficulty. At present, obstacle avoidance technology has successfully applied to unmanned surface vessel(USV) working in controlled and static environments. The obstacle avoidance algorithm applied to single USV has been very mature, but it is not fully applicable to the USVs. With the continuous breakthrough of obstacle avoidance technology, there is a lack of systematic review of dynamic obstacle avoidance of USVs. Starting from the progress of formation organization technology and dynamic obstacle avoidance technology, this paper summarizes the research progress of dynamic obstacle avoidance technology in USVs from three aspects :rule algorithm based on cluster algorithm, optimization algorithm based on model predictive control algorithm and intelligent algorithm based on reinforcement learning algorithm. Finally, the research direction and future development direction of dynamic obstacle avoidance technology in USVs are summarized and prospected.

Key words: Unmanned Surface Vessel; Formation Organization; Dynamic Obstacle Avoidance; Cluster; Model Prediction; Reinforcement Learning