Inspección robótica autónoma para reconstrucción tridimensional de grandes estructuras espaciales

Abstract

The development of large space structures, such as orbital data centers or space stations, increases the need for autonomous inspection, maintenance, and repair systems. Currently, many of these tasks rely on standardized interfaces, visual markers, or predefined sequences, solutions that lose scalability when dealing with large, diverse structures or structures not specifically designed for autonomous inspection. This work proposes a markerless autonomous inspection system based on the planning of enveloping trajectories around the structure under analysis. During these trajectories, synchronized multimodal RGB-T-D data are acquired, combining visual, thermal, and depth information. In addition, different 3D reconstruction techniques are evaluated in order to select the most suitable method for integration into the proposed test bench. The acquired data are used to generate a three-dimensional geometric reconstruction of the scene, intended to support the subsequent diagnosis of mechanical and thermal faults. The results obtained show the feasibility of the proposed approach for autonomous robotic inspection in space environments.