Stereo vision based relative state estimation for non-cooperative spacecraft with outliers

This paper proposes a theory based on stereovision to estimate the relative position and attitude between non-cooperative spacecrafts. Good performance of the scheme can be presented even with outliers in real time observation. As the information of non-cooperative spacecraft in space is not accessible, a full consideration to the target freely tumbling in space is given. Firstly, relative position and orientation equation is derived from target´s dynamics model. The extended Kalman filter (EKF) is designed to estimate the relative position and relative attitude by using the observation data provided by stereovision system. In addition, the full-dimensional state estimator is used to improve the precious of the relative position estimation. Furthermore, M-estimation is applied to improve the robustness and reliability of the previous mentioned algorithms. Simulation results demonstrate the effectiveness and convergence of the proposed algorithm. The filtering algorithms with M-estimation can effectively improve the accuracy of relative navigation for non-cooperative spacecraft when the images data is affected by outliers.