In view of the problems of slow convergence speed of traditional rapidly-exploring random tree (RRT) algorithm in complex environments, low path planning quality, and may cause problems such as low turning quality when cornering. The idea of integrating Dijkstra algorithm on the sampling basis of RRT algorithm is proposed. The improved algorithm is composed of the RRT algorithm traversing the map to from the initial undirected graph path map, and then introduces the greedy idea of the Dijkstra algorithm to search for a less expensive and better path. Under the premise of ensuring obstacle avoidance, the path obtained by the search is smoothed by using Bezier curve. After applying the improved algorithm to the simulation environment, the results show that the planning ability of the improved algorithm is better than the traditional RRT algorithm.