The situation of the post-disaster site of coal mine underground is complex, for the traditional A* algorithm of coal mine robot rescue path planning exists path insecurity, low efficiency and other problems, combined with the raster map and the characteristics of the rescue robot can be moved along any direction, multi-faceted improvement of the traditional A* algorithm. Firstly, the concept of robot size superimposed to obstacles is introduced to ensure the safety of the path. Secondly, the computational complexity is reduced by optimising the heuristic function to reduce the number of extensions that deviate from the optimal path nodes, and to reduce the redundant nodes and path transitions. Finally, a dynamic adjustment algorithm is used to round the corners of the path turns, and the resulting path is smoothed to improve the operability of the path. The optimised path with smaller path cost, shorter time consumption and fewer turns is finally obtained. Based on the improved A* algorithm, the simulation simulation test of rescue robot path planning under different sizes and different obstacle coverage is carried out using MATLAB, and the results show that compared with the traditional A* algorithm, the improved A* algorithm in rugged environment maps reduces the searching nodes by 26.51%, the computation

time by 80.42%, the path length by 31.85%, and the number of turns by 44.44%, which improves the operation of the path. reduced by 44.44%, which improves the search efficiency and smoothness of the path. The improved A* algorithm provides a theoretical basis for the rescue work of coal mine robots under complex working conditions.