Study On Visual Navigation Method For The Lunar Rover

In the second-stage projects of China's lunar exploration, the lunar rover is an important component. Navigation system is a key part of the design of a lunar rover. A lunar rover needs to complete path planning, obstacle avoidance and the scheduled tasks in an unfamiliar and unstructured environment. Therefore, to ensure that their work environment safe, the location and motion attitude should be firstly known and the information should be provided to the control system in time for the next security running to make full preparations.Visual navigation is a new method of lunar rover navigation. Visual navigation system shares a part of the equipments with vision systems, and provides a useful supplement and backup besides the conventional navigation methods of the lunar rover navigation system. This paper divides visual navigation into two parts which are stereo vision-based navigation and monocular vision-based navigation, and analyze the advantages and disadvantages of the common visual navigation algorithm. Then, the study on the visual odometry based on the stereo vision and terrain matching and lander monocular visual navigation methods is presented. At the aspect of visual navigation based on the stereo vision, the visual odometry method based on feature detection and tracking is investigated and the main factors affecting navigation precision is analyzed. Then we proposed a compact terrain matching algorithm to position the lunar rover based on the terrain from lunar rover and orbiter or lander. These two algorithms make further use of the reconstruction digital elevation map. This algorithm makes the map provided by stereo vision system can be used not only for path planning and obstacle avoidance but also can be used for visual navigation.To the monocular vision navigation, navigation method based on the single camera on lunar lander and light spots is researched. By using of the location relationship of at least four feature light spots with rectangular configuration on the lander, three-dimensional coordinates of points can be optimized. And then the position and attitude of lunar rover can be solved. This method can achieve relatively high precision within a short distance, and can be used as a backup when stereo visual navigation cameras are partly failed.