Distributed Fusion for Satellite Orbit Determination with Angles-only Measurements

This paper studies the distributed state estimation problem for noncooperative targets with angles-only measurements. We first establish the angles-only observation model of satellite formation and use the extended Kalman filter to obtain the states of noncooperative targets. To address the mild assumption on communication topology and local observability, a peer-to-peer distributed fusion is introduced, wherein each satellite interacts with its interconnected peers to attain consensus on weighted measurements and inverse-covariance matrices. Then, to reduce the error of the dynamic consensus algorithm, a consensus approach is introduced to unify the estimation results of each satellite. Moreover, the stability and convergence of the distributed Kalman Filtering are examined. Finally, simulation results show that the proposed estimation approach has better performance.