Feature-tracking based pose estimation for autonomous rendezvous and docking

Schemes for feature tracking and relative pose estimation for the final approach of rendezvous and docking procedure is proposed in this paper. Positions of the feature points were predicted by Kalman filter proposed to reduce the searching area, which effectively increased both the calculation speed and the anti-disturbance ability. Relative pose information was retrieved from solving the 3 points problem, and was taken as input of the control loop to navigate the target spacecraft to a fixed position. A vision based RVD simulation platform was built upon the on-ground semi-physical experiment platform, and experiment results showed that the proposed algorithm successfully reduced the searching error and greatly enhanced the calculation speed.