Relative State and Inertia Estimation of Unknown Tumbling Spacecraft by Stereo Vision

A novel algorithm is proposed to estimate the relative state, including position, attitude, linear velocity, angular velocity, and inertia parameters of an unknown tumbling spacecraft, by stereo vision. Feature points of the target are selected in situ, and their positions and velocities are estimated by the measurements of perspective projection and optical flow. Then, the relative attitude and angular velocity of the spacecraft are estimated by a unit quaternion method and least square method, respectively. After that, the relative position and translational velocity of the spacecraft, together with the relative positions of the detected feature points, are estimated simultaneously based on the relative translational motion model of the target by successive images. Finally, inertia parameters of the spacecraft are estimated by a quadratic optimization method based on angular momentum conservation subject to physical constraints. The performance of the newly proposed algorithm is verified by comparing with an existing case in the literature. Moreover, the performance is validated by Monte-Carlo simulations in different cases.