Motion and inertial parameter estimation of non-cooperative target on orbit using stereo vision

• Published in: Advances in space research 2020-09, Vol.66 (6), p.1475-1484
• Main Authors: Ge, Dongming, Wang, Dayi, Zou, Yuanjie, Shi, Jixin
• Format: Article
• Language: English
• Subjects: Inertial parameters; Motion estimation; Non-cooperative target; Stereo vision
• ISSN: 0273-1177; 1879-1948
• DOI: 10.1016/j.asr.2020.05.029

•Motion and parameter estimation of non-cooperative target using stereo vision.•New EKF algorithm using three non-collinear feature points of target as observation.•The information obtained includes the target motion state and inertial parameters.•The convergence of the estimation algorithm is verified by mathematical simulation.•Good estimation performance with an acceptable accuracy is achieved by experiment. In the on-orbit servicing missions, autonomous close proximity operations require knowledge of the target’s motion and inertial parameters in order to safely interact with the target. In particular, the motion and parameter estimation of an uncooperative target is a challenging task because of the lack of some prior information about the target in unfamiliar environments. In this paper, a novel method is developed for an accurate estimation of the motion and inertial parameters of an unknown target using stereo vision measurements only. Instead of using relative position and pose as observation information, this paper chooses three non-collinear feature points on the target as estimation measurements. This treatment allows us to estimate the principal axis of inertia of the target directly, and avoid estimating the quaternion between the target measurement coordinate system and the principal axis of inertia coordinate system, which is a bilinear problem, and is difficult to obtain global convergence. Based on the relative kinematics and dynamics equations of the target, the state equation and observation equation are derived, and the Extended Kalman Filter (EKF) is designed. As a result, the developed algorithm realizes the estimation of the complete unknown information of the target, including relative position, relative velocity, attitude quaternion, angular velocity, the direction of the principal axis of inertia, inertia ratio, and the position of center of mass. Numerical simulations and experimental results verify the convergence and effectiveness of the proposed filter estimation method.