Relative state estimation of space debris based on the stereo-vision measurement system

The increasing number of space debris objects threatens the safety of space assets. A number of proposed solutions to the debris problem involve dedicated debris removal satellites to reposition or de-orbit uncontrolled objects, using grappling devices, tethers, or electrostatic forces. In order to conduct proximity operations necessary to rendezvous and de-orbit space debris, accurate relative navigation is required. The space-based stereo-vision measurement is an effective solution for relative navigation. However, when tracking space debris with large separation distances, the relative state estimate can become less accurate. This work uses Lawden’s equations as the relative dynamical model for the elliptical reference orbit and analyzes the uncertainty of the relative state estimate. An innovative approach is proposed to compute the relative position PDF by using Monte Carlo samples and the Gaussian mixture model method. The initial relative velocity PDF is computed using the unscented transform with the PDFs for the first two initial positions and linearized relative dynamics. The Gaussian mixture unscented Kalman filter is employed to refine the estimate based on subsequent measurements. Simulation results validate the efficacy of the proposed method.