Visual navigation is the key technology to enable autonomy and meet the requirement of pin-point landing in future asteroid exploration missions. Various innovative vision-based methods, with respect to multi information fusion, landmark extraction and matching, etc., have been proposed in the last decades, but still more or less encountered some critical issues (e.g. efficiency and precision) in practical. This paper focuses on asteroid landing and presents a visual navigation method based on Gauss-Newton iterative algorithm to minimize the reprojection errors between 3-D landmarks and the corresponding 2-D projections on the imaging plane. First of all, the Lie algebras se (3) of Euclidean transformation matrix is used as the probe’s state, which leads the optimization problem to an unconstrained one, and based which, the landmark measurement model is constructed and followed with the observability and navigation error analysis. Then, an effective and very fast sub-optimal landmark selection method—Maximize the Sum of Distances (MSD) between landmarks, based on the error analysis, is designed to release the computational burden of state estimation. Finally, numerical simulations are applied and the simulation results validate the effectiveness of the proposed navigation algorithm and demonstrate its high efficiency and precision.
Visual Navigation Based on Nonlinear Optimization Method for Asteroid Landing
Lect. Notes Electrical Eng.
2021-10-30
13 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Landmark robust selection for asteroid landing visual navigation
Elsevier | 2023
|Autonomous Navigation for Binary Asteroid Landing
AIAA | 2024
|