VARIABLE RESOLUTION QUADTREE MAPPING FOR PLANETARY LANDING USING PLANAR ELEMENTS

We present a terrain mapping technique for autonomous spacecraft landing on planetary bodies. An estimate of the terrain is formed without any a priori knowledge using generic range and bearing measurements. The map is represented with the quadtree, a variable resolution discretization. This data structure allows high resolution representations in areas with complex terrain features and low resolution representation in areas with sparse measurements or simple terrain features. Each quadtree cell contains a planar estimate of its terrain, these planar estimates are initialized with a Maximum Likelihood estimate and refined with an Extended Kalman Filter. The planar estimation strategy is validated through Monte Carlo and the mapping technique is demonstrated with an planetary landing simulation based on real flight tests conducted as part of NASA’s ALHAT program.