SPACE OBJECT TRACKING IN THE PRESENCE OF ATTITUDE-DEPENDENT SOLAR RADIATION PRESSURE EFFECTS

The tracking of space objects (SOs) is often accomplished via measurement systems such as optical telescopes, where the observations occur infrequently, leading to long periods of time in which the SO uncertainty must be propagated. Given that the uncertainty associated with an SO is described by a probability density function (pdf), the infrequent measurements and subsequent long arcs of propagation lead to difficulties in reacquisition due to inaccurate methods for the propagation of the pdf. The standard methods for propagation of the pdf are typically those of the extended Kalman filter (EKF) or unscented Kalman filter (UKF) which make use of only the first two moments of the pdf, thereby limiting their ability to accurately describe the actual pdf. This work examines an improved propagation scheme which allows for the pdf to be represented by a Gaussian mixture model (GMM) that is adapted online via splitting of the GMM components based on the detection of nonlinearity during the propagation. In doing so, the GMM approximation adaptively includes additional components as nonlinearity is encountered and can therefore be used to more accurately approximate the pdf. The improved representation of the uncertainty region of the SO can then be used to more consistently reacquire SOs, thereby improving the overall tracking of space objects. It is shown (via simulation) that the improved propagation scheme leads to a more realistic and accurate representation of the region of uncertainty. Furthermore, when the more accurately propagated uncertainty is fused with incoming measurement data, it is shown that the resultant updated uncertainty more accurately represents the true state and yields a smaller region of uncertainty.