In order to achieve the goal of the effective path planning and trajectory tracking of UAV (Unmanned Aerial Vehicle), one approach based on the RRT (Rapidly Exploring Random Trees) algorithm proposes a path planning algorithm for UAV in the environment of electric towers. Under the combination of the dynamics characteristics of UAV and the randomness of RRT, the algorithm with constraint of the maximum path angle and path height plans a path which gradually approximates to the target point and overcomes the failure problem of RRT in the narrow space planning. The simulation experiment is established in the ROS (Robot Operating System)-Gazebo and verifies the practicability of the algorithm.