Feature-based Optical Navigation for Lunar Landings

Abstract

For space exploration, reducing the size of landing ellipses, increases the access to areas of scientific interest. This can be achieved with improved GNC systems. Since the exclusive use of IMUs is insufficient for Pinpoint Landing, optical navigation is being explored as a solution. The method developed in this work is based on image features matched between a pre-built database and the landing images captured in real-time. The algorithm was tested with data from DLR's TRON laboratory, namely images of a terrain model representative of the lunar surface and the associated ground-truth. Provided sufficient image area, the navigation solution averaged around 1% of the Line Of Sight (LOS) distance and had a maximum value of around 3% of the LOS distance (apart from clear outliers).