Abstract: The path planned by traditional jump point search (JPS) algorithm in complex environment has many expansion nodes, uneven path, poor security and difficulty to follow the optimal path. A fusion method based on improved JPS algorithm and dynamic window approach (DWA) is proposed. Firstly, JPS algorithm is improved to search the path alternately from the starting point and the target point to improve the efficiency of AGV (automated guided vehicle) path search. Secondly, the environmental obstacle rate optimization heuristic function is introduced to enhance the safety and purpose of AGV path search. Thirdly, the improved Floyd algorithm is used to process the obtained path to ensure that the shortest path is also a safe path. Then, the dynamic tangential point adjustment method is used to smooth the turning point to make the path conform to the dynamic characteristics of AGV. Finally, in the evaluation function of DWA algorithm, the known and unknown obstacles are classified, and the azimuth evaluation function is improved, so that the dynamic path planning can take into account the global optimization of the path and the real-time obstacle avoidance ability. In order to verify the effectiveness of the algorithm, comparative simulation experiments are carried out in raster maps of different complexity. The experimental results show that the paths planned by the improved JPS algorithm have an average reduction of 23.3% in the number of expanded nodes, an average reduction of 45.8% in the turning angle, an average reduction of 48.1% in path search time, and an average reduction of 5.7% in path length compared to the traditional JPS algorithm. The paths planned by the proposed fusion algorithm have an average reduction of 6.1% in path search time and an average reduction of 0.9% in path length compared to the traditional fusion algorithm.