The path planning of unmanned surface vehicle is the key to ensure that unmanned vehicle can travel safely and complete the task efficiently. Aiming at the path planning of unmanned surface vehicle in complex environment, In this paper, the improved A* algorithm is adopted in global path planning to remove potential collision paths and redundant points at corners, reducing 29.41% of path points, 9.65% of path length and 14.39% of planning time. In order to solve the local path planning of unmanned surface vehicle in dynamic environment, multi-objective particle swarm optimization and artificial potential field method are adopted. To prevent collision between paths and obstacles, obstacle avoidance constraints are constructed.At the same time, the particle spacing constraint is constructed to make the path smooth. Combined with the multi-objective particle swarm optimization algorithm, the safety distance of unmanned surface vehicle is introduced to optimize the weight coefficient and the repulsion gain coefficient in the artificial potential field method. Secondly, the global path is further optimized, considering that if there are no obstacles between particles, the length of the path is further shortened, and the efficiency is improved. Through MATLAB simulation, the results show that the improved algorithm can flexibly avoid static and dynamic obstacles, taking into account both safety and speed.