3D terrain feature matching based navigation for lunar soft landing

A 3D terrain feature matching based autonomous navigation method for lunar pinpoint landing is presented in this paper. Firstly, 3D terrain features are detected by searching local maximum of the digital elevation map's local variance. Secondly, based on the invariants of relative distance and angle, a vote method is used to match detected terrain features with the feature database of global digital elevation map. Thirdly, for uncertainties of measurement noise statistics in pose estimation, the Sage-Husa noise statistics estimator is fused with an iterative Kalman Filter to provide accurate estimation of spacecraft position and attitude. Simulation results are given to show feasibility of the proposed navigation scheme.