Automated state and dynamics estimation in dynamically mismodeled systems with information from optimal control policies

While observation sets for an individual object in orbit can be quite data-sparse, the sheer number of objects in orbit makes the tracking problem as a whole data-rich. As such it is infeasible for humans to process these measurements manually. While orbit determination procedures are largely automated the resulting solutions can be quite poor when the assumed dynamics are flawed due to mismodeled perturbations or unmodeled active maneuvers. This paper presents an automated version of the Optimal Control Based Estimator (OCBE) as applied to problems in the field of Space Situational Awareness. The OCBE algorithm supplements a state and dynamic estimation process with information from optimal control policies to create an estimator that is robust to dynamic mismodeling. This estimator is made adaptive through a maneuver detection based approach resulting in an algorithm that automatically estimates the system's state, detects dynamic mismodeling, and reconstructs that mismodeling. Along with a derivation and discussion of this algorithm, numerical simulations are provided to demonstrate this algorithm's ability to accurately track a spacecraft with mismodeled dynamics while simultaneously detecting and reconstructing those mismodelings.