Abstract: A planetary rover is a spacecraft that can move on the surface of a solid planet, performing tasks such as exploration, sampling, and transportation. It is one of the important methods of deep space exploration. The autonomous path planning method of the planetary rover is the key core part of the rover navigation technology and it is an important support for ensuring the safety and efficiency of the rover’s exploration tasks. Compared with the autonomous path planning methods of ground mobile robots, due to the influence of factors such as the special environment of the outer planet and the communication delay and bandwidth limitation, the rover needs to find a safe, efficient, and feasible driving path autonomously without human intervention, based on incomplete environmental information and uncertain position information, while considering factors such as terrain and motion constraints. This makes the autonomous path planning method of the rover face more complexity, uncertainty and unknownness. Therefore, the relevant research situation is analyzed and sorted out from two aspects in this paper: the perception map construction method and the autonomous path planning method of the rover. First, the method and research progress of generating digital elevation map models based on stereo vision information are introduced, especially the stereo feature matching methods are divided into region-based matching algorithms and feature-based matching algorithms for sorting and analysis. Secondly, the basic principles and application situations of the autonomous path planning method of the rover are classified and sorted out, and the autonomous path planning method of the rover is divided into cost-based path planning method and machine learning-based path planning method for technical analysis. In the cost-based path planning method part, the autonomous path planning method based on the candidate arc is analyzed, the iteration of the autonomous path planning method of the launched rover is sorted out, and the autonomous path planning methods based on A*, RRT*, and FFM are classified and analyzed. In the machine learning-based path planning method part, the application of machine learning in the autonomous path planning of the rover is sorted out from two aspects: end-to-end path planning method and machine learning-assisted path planning method. Finally, based on the technical analysis, the future research direction of the autonomous path planning method of the rover is discussed and prospected from four aspects: enhancing the perception ability of the rover, improving the candidate arc, exploring the combination of multiple path planning methods, and strengthening the application of machine learning.