Aiming at the randomness of the random sampling of the free space of the path planning and the no solution problem of the possible track in the narrow channel

a path planning algorithm for improving probabilistic road map（ PRM） is proposed. By using the boundary point of the obstacle as the sampling point and establishing the optimal feasible area for the free space of the grid map

the algorithm reduces the dispersion of the random sampling points of the traditional PRM algorithm and makes the path search clear

to improve the time and space utilization of the algorithm. In order to verify the effectiveness of the algorithm

the MATLAB simulation platform is used to establish the two-dimensional（ 2 D） and three-dimensional（ 3 D） space of the UAV

and the path planning comparison experiment is carried out. The experimental results show that in the 3 D space

the improved PRM algorithm is reduced by 2. 469 % ～ 5. 721 % and the length of path is reduced by 0. 89 % ～ 1. 54 % compared with the traditional PRM algorithm.