Non-Recovering Field-of-View Imaging-Based SLAM for Lava Tubes Exploration
Résumé
Visual Odometry (VO) has emerged as the go-to navigation solution for space missions, providing a reliable and efficient means of estimating the motion and state of unmanned vehicles. Recently, the exploration of potential extraterrestrial lava tubes has garnered significant attention within space agencies, presenting a captivating challenge in applied robotics. Navigating through these unique environments demands a wide Field-of-View (FoV) to efficiently catch information on the majority of the surroundings; however, most of rovers are equipped with narrow FoV stereo-vision configurations for direct mapping, traversability estimation and navigation. Conventional earth-based approaches to expand the FoV often prove unsuitable for space applications due to power, processing, or technical constraints. In this paper, we propose an innovative solution: indirect bi-monocular VO with a non-recovering FoV in order to make the most effective use of available camera pixels. Leveraging sliding-window optimisation techniques, we aim to overcome the inherent difficulties in accurately estimating scale in displacement and environment. Our approach paves the way for achieving robust scale-aware navigation in a non-overlapping camera configuration, opening new frontiers for space exploration methodologies in extraterrestrial lava tubes.
Domaines
Robotique [cs.RO]
Origine : Fichiers produits par l'(les) auteur(s)