3D Model-based Visual tracking for Space Autonomous Rendezvous

International audienceFor a space rendezvous mission, autonomy imposes stringent performance requirements regarding navigation, for the guidance and control to be managed. For the final phase of the rendezvous, a vision-based navigation, using a monocular camera, can be a solution. Markerless 3D model-based tracking has been studied and tested on a mock-up of a telecommunication satellite. Tests on typical approach scenarios have shown satisfactory results, in terms of precision of the pose estimation and computational costs. Quantitative tests have also been carried out to determine the robustness of the algorithm to inter frame motions, orientation, illumination conditions and initialization uncertainty. The tracking algorithm has also been successfully implemented in a closed loop chain for visual servoing