ESA GNC Conference Papers Repository
3D model-based relative pose estimation for rendezvous and docking using edge features
Automated rendezvous and docking is one critical research area for space servicing and human/robotic space exploration. This paper presents a 3D model-based relative pose estimation process, which is one crucial step towards automated rendezvous and docking with non-cooperative targets. The pose estimation is based solely on a 3D CAD model, infrared images and their extracted edges. Parameters of the special Euclidean group's Lie-algebra are determined via an iterative re-weighted least squares algorithm. The resulting parameters define the position and attitude of the target. For creating the results, real infrared image sequences were used. They were recorded during the rendezvous and docking of the ATV-5 with the ISS. Experimental results demonstrate the successful pose estimation during the whole approaching process, even during terrestrial eclipses.