Investigation of variable discretization resolution for CD-EKFs in space object tracking

Abstract

Space object (satellite or space debris) tracking has been identified as a key component of Space Situational Awareness. Space object tracking is a continuous-discrete filtering problem. Conventional extended Kahnan filter (EKF) and un-scented Kalman filter (UKF) methods are formulated for discrete-discrete filtering problems. New versions of the EKF have been engineered recently for continuous-discrete filtering problems. We first discuss the dynamic and observation model of the space object tracking problem on which we later on test 5 filters in the CD-EKF framework. Our results show that for the space object tracking problem with radar, solutions which discretize the Langevin equation give results comparable to Moment-Matching based EKFs at very high discretization resolutions.