Abstract
Abstract. Amidst the exponential increase in space debris, Active Debris Removal and On-Orbit Servicing missions have gained paramount importance. The success of such endeavours pivots on the comprehension of the target's geometry and dynamic conditions, highlighting the indispensable nature of target inspection. In recent times, a variety of inspection missions have employed satellites of different classes. In this context, small satellites like CubeSats have emerged as a reasonable solution, due to cost-effectiveness and rapid development capabilities. This study explores a potential inspection mission utilizing a CubeSat to evaluate a non-cooperative target, where information on relative pose are not known a priori. The aim is to retrieve essential data for executing Close Proximity Operations safely. Manoeuvring around the target, the CubeSat captures two-dimensional (2D) RGB images from multiple angles to then reconstruct its three-dimensional (3D) geometry. Optical cameras are preferred during the inspection phase due to their cost-efficiency and low power requirements, different from other technologies like laser imaging. An experimental setup is designed and built to generate a dataset of 2D images simulating in orbit conditions. Standard computer vision algorithms are employed to perform the 3D reconstruction and Artificial Intelligence used to reconstruct the scene.
Publisher
Materials Research Forum LLC