Combined optimal control and state estimation for the purposes of maneuver detection and reconstruction

A new type of estimator that incorporates optimal control and outputs a control policy is developed and analyzed in this study. The estimator is developed in a similar manner to a Kalman Algorithm with an almost identical form, but has additional properties for more accurate tracking, maneuver detection, and maneuver reconstruction. Unlike the Kalman Algorithm, this estimator frees up the initial state, which results in an algorithm that decouples a priori state uncertainty and dynamics uncertainty. The dynamic uncertainty inflates the state covariance in an automatic fashion that prevents filter saturation. The algorithm also outputs control estimates that may be used to both identify the presence of mismodeled dynamics and quantify those mismodeled dynamics. An example application demonstrates the maneuver detection and reconstruction properties of this estimator in an orbit determination problem where the dynamics are mismodeled.