Stereo Vision Based Motion Estimation for Lunar Rover Navigation

Accurate navigation autonomously on uneven terrain is a very important technique for lunar rover to execute long-range exploration on lunar surface. This paper presents a stereo vision based algorithm for lunar rover autonomous navigation that can enable for precision localization. Our techniques mainly consist of image processing and motion estimation. To improve the performance of motion estimation, robust linear motion estimation is executed to reject the outliers of image processing and estimate the motion initially, then, Levenberg-Marquardt nonlinear estimation is executed to estimate the motion precisely. The result of our autonomous navigation algorithm is an estimation of attitude and position, which can be passed directly to path planning and motion control system