An improved A* algorithm based path planning method for unmanned underground loadhauldump (LHD) was proposed and verified. According to the articulated structure of the underground LHD, the nodes were expanded by the articulated angle, which makes the expended nodes match the trajectory characteristic of the LHD. In order to avoid collision, collision threat cost was designed and added into the evaluation function, which makes the navigation path exclude against the walls. Simulation and comparison results between the improved A* algorithm and tranditional one show that the improved A* algorithm is more efficient, and the shortest path without collision can be obtained when the weighting coefficient is 0.2. Moreover, a path planning and tracking test is carried out in lab environment. It is shown that the tracking precision can keep in a small range and the LHD can run without collision when the navigation path is given by the proposed algorithm, which proves the effectiveness and feasibility of the proposed algorithm.