Nonlinear State and Mass Property Estimation via Square-Root Unscented Kalman Filter on TSE(3) with Windowing Function

Mass property estimation, encompassing mass, center of mass, and moment of inertia, represents a fundamental yet inherently challenging problem in spacecraft autonomy and astrodynamics. Accurate estimation of these properties enables significant reductions in operational overhead and enhances spacecraft control performance. This paper builds upon previous work [1–3] by proposing a robust dual square-root unscented Kalman filter framework defined on the tangent bundle of the special Euclidean group TSE(3) explicitly designed for nonlinear mass property estimation. The primary contributions of this work are as follows: (1) a novel windowing strategy that improves estimator stability and convergence under large initial uncertainties; (2) incorporation of a numerical propagation technique, specifically adapted for use on SE(3), to enhance dynamic accuracy; and (3) development of a systematic open-loop thruster excitation sequence that minimizes actuation demands while ensuring adequate excitation for reliable parameter identification. Comprehensive Monte Carlo simulations validate the proposed approach, demonstrating rapid convergence, robustness to substantial initial estimation errors, and resilience to sensor noise characteristics typical of commercially available spacecraft instrumentation.