OPTICAL ONLY CO-ESTIMATION AND MANEUVER CLASSIFICATION IN THE CISLUNAR REGIME

Traditional concerns regarding spacecraft estimation are exacerbated in cislunar space due to highly nonlinear dynamics, limited observational information, and increasingly overloaded Earth-based resources. These concerns motivate the investigation of space-based navigation and observation capabilities which lend themselves to autonomy. This paper demonstrates that relative optical measurements provide co-estimation of a chaser and target spacecraft, while simultaneously classifying unknown maneuvers. First, linear observability analysis proves that relative optical measurements produce a fully observable dual spacecraft system. Subsequently, the Optimal Control Based Estimator is modified with the Unscented Transformation to account for the strongly nonlinear dynamics and leveraged to generate a state solution along with an estimated control policy. Finally, the estimated control policy is employed to identify and classify unknown maneuvers from the target spacecraft via machine learning. Collectively, this methodologydirectly enables a chaser craft to autonomously navigate to a target and provides a basis for further research into autonomous operations in cislunar space.