Space-Object Active Control Mode Inference Using Light Curve Inversion

Abstract

Several candidate methods for classifying an agile space-object active control mode are tested using simulated light curve data and a Rao–Blackwellized particle filter. The first measure to discriminate the space-object control mode is a measurement dissimilarity metric, defined as the time integral of the error between the estimated target space-object sensor boresight and the line-of-sight vector to each hypothesized subject. The second measure quantifies the “pointing quality” using the multivariate Gaussian mixture model analog to the Mahalanobis distance, which is computed using the estimated multivariate body angular velocity distributions. It is shown how additional information from the space-object shape model can be combined with radiometric first principles to establish a tracking error threshold between the target space object and the hypothesized subject. Finally, the body angular velocity estimates are used to compute the mass-specific rotational angular momentum and mass-specific rotational kinetic-energy analogs. These analogs are coupled with statistical inference techniques to classify the active control modes of agile space objects.