Matching Algorithms for Target Localization in Pinpoint Landing

Pinpoint landing identifies the capability to land on a planet, e.g. Mars or the Moon, with a position error lower than few tens of meters. To reach this challenging aim different technologies are involved. Looking deeper at the area of the terrain relative navigation and strategies, this work focuses in particular on the development and analysis of two matching algorithms designed to initialize the landing sequence. The primary purpose of the matching algorithms is to provide an initial estimate of the spacecraft pose with respect to the landing target, encompassing both position and attitude. This estimation serves as a crucial input for initializing the guidance and navigation system during the final descent phase towards the established landmark. To achieve this, during the descent phase a Landing Vision System (LVS) mounted on the spacecraft captures multiple images of the planetary surface and various landmark features- such as radius and relative coordinates- are extracted. The matching algorithm then compares these obtained features with a preloaded map of a broader region encompassing the captured image. For the objective of this dissertation, two matching algorithms have been developed and analyzed. The first algorithm is based on the construction of triangular configurations, leveraging the geometric similarity of these structures for feature matching, whilst the second one seeks the transformation that optimally aligns the extracted features with those present in the reference map. Furthermore, a Monte Carlo analysis was conducted to assess the performance of the aforementioned methods, accounting for errors affecting both attitude and altitude. The results demonstrate that both algorithms exhibit robustness, even under challenging conditions. While the first algorithm is more resilient to errors, it entails a higher computational cost. Conversely, the second algorithm is approximately ten to twenty times faster but exhibits greater susceptibility to inaccuracies and errors. In the conclusions, this thesis describes the manners and constraints for the applicability of such typologies of algorithms in the frame of the future exploration missions, manned and unmanned, where position accuracy at landing will become more and more crucial.