Abstract: In view of the robot at the bottom of the sea under complicated environment, the traditional PSO path planning algorithm for continuous wave problems, lead to a larger submarine robot speed and position of disturbance, get rid of the obstacles path optimization results, there is a big deviation of an improved PSO underwater robot path out of optimization method under complicated environment, into the speed-position search model in the traditional algorithm, using for high-dimensional function and cognitive function obstacle constraints limit, at various stages of the robot obstacle from the adjust value of cognitive factor and environmental factor, get the best out of obstacles path, make the underwater robot can efficiently get rid of the obstacles. In particle number is 200 N, particle dimension to 10, the maximum number of iterations for 800 experiments, this algorithm make the results of the path planning with the optimal path planning of has high compatibility, error control within 3.5%, the obstacles to get rid of the effect is more obvious, has good application value.