A Nonlinear Variable Dimension estimator for maneuvering spacecraft tracking via the Unscented Kalman Filter

A nonlinear variable dimension (NVD) adaptive estimator was proposed for tracking maneuvering targets in the spaceflight setting. To address the uncertainty of the nonlinear system dynamics, a single unscented Kalman filter is used with two state models: a nonmaneuvering model for the two body problem and a higher-order maneuvering model that takes maneuver acceleration as a deterministic but unknown parameter to be estimated. A maneuver detector based on statistical tests is developed for directing the system model switch. The NVD estimator proposed in this paper is suitable for tracking nonlinear systems with deterministic mode uncertainties. It is potentially a cost-effective online method for real-time spacecraft tracking applications.