The study of hybrid vision-based navigation for lunar rover in virtual environment

Autonomous navigation is a very important function for lunar rover, which plans a safe and efficient path by rapid detection and recognition of obstacles in a negotiable region. Simulation and verification of the navigation of lunar rover are very difficult, cost on the moon. The paper researches and presents a fusion monocular, binocular visual navigation of the hybrid method on virtual environment. It gives a monocular motion image vision navigation method to give a path planning for micro-scene, raw planning and a binocular stereo visual navigation method to avoid obstacles in close scene. Meanwhile we combine vision-based navigation into virtual simulation system based on a full-digital lunar terrain, integrated with kinematics and dynamics module. An autonomous navigation simulation module has been integrated in this prototype system, which was proved by the simulation results that the synthetic simulation and visualizing analysis system are established in the system; and the system can provide efficient support for research on the autonomous navigation of lunar rover.