Experimental simulation of satellite relative navigation using computer vision

This paper presents a general description of a mechatronics experimental testbed developed to simulate relative autonomous navigation between satellites using computer vision. Since the experimental tests of relative navigation is not feasible in real systems for the costs and risks that implies, it was decided to construct an experimental platform that simulates the spatial mission scenario. The mission to simulate deals with the recognition and inspection of a satellite by means of another autonomous satellite (Chaser), which approaches its target and verify its actual state. Both relative navigation of the chaser and inspection of the target are based on computer vision. The experimental testbed will be used for the simulations of service missions of satellites to test the performance of the developed vision and navigation algorithms and techniques under space dynamics laws. The system uses a 3D model of the object to achieve the estimation of its position and orientation (pose), and to perform tracking by means of images sequences. The pose estimation and feature correspondence determination are to be solved by algorithms that are capable of achieving them simultaneously. The paper presents and discusses the initial results of the two algorithms that are being used in the simulation testbed.