A drone needs to plan flight paths to the satisfaction of constraints according to the external environment in the process of mission execution. Aimed at the problems that local minimum is in existence in the traditional artificial potential field (APF) planning paths, and the failure of path planning is unable to get out of the trap, narrow path oscillation and other defects, a complex obstacle escape path planning method is proposed based on random tree potential field (RTPF). First, a discrete drone environment map model is established to generate a random tree path in each environment map by using rapidly exploring random trees (RRT), and a random tree potential field function is designed to construct the potential field for the path nodes. And then, by reasonably designing the parameters and introduction timing of random tree potential field, the drone is guided to escape from the potential field environment in the defective situation, solving the defective problem of the traditional artificial potential field method. Finally, the algorithm in the paper is simulated and analyzed with the comparison algorithms. The results show that com pared with the fast search random tree algorithm and the traditional artificial potential field method, the algorithm in the paper completes the path planning in all obstacle cases and shortens the length of planned path, further improving the safety and effectiveness of path planning.