In the unstructured scene with multiple obstacles, the traditional hybrid A* algorithm has the problems of low computational efficiency and poor path smoothness. For these problems, this paper proposes a hybrid A* path planning method based on the density-based clustering (DBSCAN) and the dichotomy. Firstly, based on the DBSCAN algorithm, an obstacle clustering method is designed to simplify the multi-obstacle unstructured scene, so as to avoid invalid node expansion of the hybrid A* algorithm near the U-shaped obstacle group, and to effectively improve the efficiency of the algorithm. Then, a dichotomy-based state node expansion strategy is proposed, which can search a smoother path without significantly increasing the computational complexity of the hybrid A* algorithm. Finally, simulation is performed on MATLAB. The results show that in the multi-obstacle unstructured scene, the improved hybrid A* algorithm proposed in this paper can significantly improve the computational efficiency and the path smoothness.