Terrain Relative Navigation in the Permanently Shadowed Regions at the Lunar Poles with DEM Data and SIFT Algorithm

China's Chang'e-7 mission (CE-7) plans to conduct in-suit measurements in permanently shadowed regions (PSRs) at lunar south pole. The poor lighting conditions near the lunar poles makes optical images insufficient for terrain-relative navigation (TRN). Digital elevation model (DEM) obtained by the lunar orbiter laser altimeter (LOLA) onboard the lunar reconnaissance orbiter (LRO) is suitable for TRN in PSR. By matching the real-time DEM data acquired by the laser imaging sensor on the flyer with the geocoded LOLA DEM data, the position of the mini-flyer can be estimated. In this paper, scale-invariant feature transform (SIFT) algorithm is used to acquire matched points, which can be used to calculate rotation and scaling matrix to determine the position and flying direction of the flyer. The images of the surface slopes derived from the LOLA DEM data with high resolution are taken as the real-time images from the mini-flyer, and the images with low resolutions are taken as the reference images. SIFT algorithm performs well under different resolutions and rotation angles. The results indicate that the matching error between reference images and real-time images is small, making it suitable for TRN in lunar PSR. This paper also conducts registration experiments between SAR images and DEM images. The results indicate that the registration performance between SAR images and DEM images is not as good as that of DEM images with different resolutions.