Stereo-Vision-Based Relative Pose Estimation for the Rendezvous and Docking of Noncooperative Satellites

Autonomous on-orbit servicing is expected to play an important role in future space activities. Acquiring the relative pose information and inertial parameters of target is one of the key technologies for autonomous capturing. In this paper, an estimation method of relative pose based on stereo vision is presented for the final phase of the rendezvous and docking of noncooperative satellites. The proposed estimation method utilizes the sparse stereo vision algorithm instead of the dense stereo algorithm. The method consists of three parts: (1) body frame reestablishment, which establishes the body-fixed frame for the target satellite using the natural features on the surface and measures the relative attitude based on TRIAD and QUEST; (2) translational parameter estimation, which designs a standard Kalman filter to estimate the translational states and the location of mass center; (3) rotational parameter estimation, which designs an extended Kalman filter and an unscented Kalman filter, respectively, to estimate the rotational states and all the moment-of-inertia ratios. Compared to the dense stereo algorithm, the proposed method can avoid degeneracy when the target has a high degree of axial symmetry and reduce the number of sensors. The validity of the proposed method is verified by numerical simulations.