Abstract: Aiming at the problems that A* algorithm cannot deal with unknown obstacles underground, DWA algorithm has redundant sections and is easy to fall into traps, and DWA combined with global path planning algorithm has large calculation, a hierarchical smooth optimization bidirectional A* guided DWA(HSTA*-G-DWA)path planning method for coal mine rescue robot was proposed. Firstly, the dynamic weighting factor and normalization factor are introduced into the cost function of bidirectional A * algorithm, and the collision constraint function is constructed to eliminate the search of unrelated extension nodes in the path search process, so as to improve the efficiency and security of path search. Secondly, the hierarchical smoothing optimization strategy is used to eliminate redundant points and turning points in the path, and reduce the number of path points and path length. After that, the path is further smoothed by cubic B-spline curve. Thirdly, if the Coal mine rescue robot detects an unknown obstacle, it uses the global path to guide DWA to achieve local dynamic obstacle avoidance, otherwise it tracks the initial global path. Finally, the experimental results demonstrate that the proposed algorithm enhances real-time pathfinding efficiency while ensuring global optimization, as well as improving trajectory security and smoothness.