According to its planning scope, path planning for unmanned surface vehicle (USV) can be divided into global and local path planning. Many scholars have improved the classic algorithms, including grids method, visibility graph method, A* algorithm and artificial potential field method (APF), But the global planning algorithm still has outstanding problems such as long calculation time and large computational overhead in large task space, local planning algorithms usually ignore the global optimal constraints. Aiming at the problem of dynamic path planning of environmental monitoring USV under complicated offshore navigation conditions, based on the idea of bi-level planning, a hybrid algorithm which combines global and local path planning is proposed. This paper first proposes an improved Particle Swarm Optimization (PSO) for global path planning according to the given information about marine environment, and introduces Opposition-based Learning (OBL) and improves the inertia weight as well as search step size to effectively avoid the precocity of PSO. Then on the basis of global optimized path and sensor information, the improved Artificial Potential Field (APF) algorithm is adopted for local dynamic obstacle avoidance, so as to solve the local minimum problem. The results of simulation indicated that the improved PSO can effectively avoid the precocity of particles and enhance the optimization capability and stability of the PSO; the improved APF would not be restricted by local minimum point and achieve dynamic obstacle avoidance under the constraints of global optimization path. Therefore, the combination of these two algorithms can effectively solve the problems of path optimization and dynamic obstacle avoidance for environment monitoring USV when it is executing missions in complex offshore areas.

根据其规划范围，无人水面飞行器（USV）的路径规划可分为全局和局部路径规划。许多学者对经典算法进行了改进，包括网格法，能见度图法，A *算法和人工势场法（APF），但全局规划算法仍然存在计算时间长，任务空间大，计算量大的突出问题。 ，本地规划算法通常会忽略全局最优约束。针对复杂海上航行条件下USV环境监测的动态路径规划问题，基于双层规划的思想，提出了一种结合全局和局部路径规划的混合算法。本文首先根据给定的海洋环境信息，提出了一种用于全局路径规划的改进的粒子群算法（PSO），并引入了基于对立的学习（OBL），并改善了惯性权重和搜索步长，从而有效避免了早熟PSO。然后在全局最优路径和传感器信息的基础上，采用改进的人工势场算法来避免局部动态障碍，从而解决了局部最小问题。仿真结果表明，改进后的粒子群优化算法可以有效避免粒子的早熟现象，提高粒子群优化算法的优化能力和稳定性。改进的APF不受局部最小点的限制，在全局优化路径的约束下实现了动态避障。因此，