Abstract: In view of problems of overshoot, collision, discontinuity and unsmoothness in the trajectory generated by motion planning of existing underground mobile robot, a motion planning method of underground mobile robot was proposed, which consists of path planning, trajectory generation and trajectory optimization. Path planning is realized by A* algorithm based on graph search, and a passable global path is quickly planned as initial value of path generation through three steps of cycle iteration, namely start search, path sorting and continue search. Trajectory generation is realized by constructing objective function based on Minimum Snap and applying equality constraint.Trajectory optimization is achieved by applying inequality constraints on the basis of trajectory generation: by adjusting time allocation and building inequality constraints on Corridor trajectory planning, the overshoot phenomenon that occurs during solution process for Minimum Snap trajectory generation is resolved and the entire trajectory itself is constrained to avoid collisions; by introducing the harmonic function Bezier Curve, the trajectory optimization problem of the Minimum Snap based on Bezier Curve is constructed, which makes the solution of the high-order objective function simple and efficient, and finally generates a motion trajectory which is suitable for the underground mobile robot with minimum energy loss, continuous, smooth, collision free and executable.In the Matlab simulation environment, the random map is designed, and the optimal trajectory planning results including time allocation, location planning, speed planning and acceleration planning are generated.The experimental results verify the correctness and effectiveness of the motion planning method.