Path planning for lunar rover involves defining the optimal route across the moon’s surface, typically considering factors such as terrain, obstacles, and target locations, to ensure the rover completes its mission efficiently and safely. This paper presents a two-stage lunar rover path planning method integrating global routing with local obstacle avoidance. This method utilizes sensor data from unmanned rovers and destination coordinates for path planning. Initially, point cloud data from various sensors are used to construct a high-resolution digital elevation map (DEM) of the local area by gridding. This detailed DEM is then compressed to a low-resolution DEM to reduce computational load. Subsequently, the local maps are merged to create a global map for global path planning, ensuring information preservation and memory reuse. Lastly, the optimal forward path is selected on the local map during each iteration by considering terrain information, target location, and global path. Our simulation experiments demonstrate that the proposed method is more efficient and accurate compared to single-stage global path planning or local obstacle avoidance methods.