Navigation With Shape Estimation During Small Body Approach

A Gaussian Process Sequential Filter (GPSF) is implemented for closed-loop mission design for approaching and mapping small bodies. The small bodies are modeled using a Gaussian Process (GP), where basis nodes and radii provide a continuous, differentiable model of the entire shape. With optical silhouette-based measurements, the GPSF simultaneously estimates the spacecraft’s relative position and velocity along with the small body’s pole orientation, spin, gravitational parameter, and shape. The optical imagery is subject to attitude knowledge errors. The GPSF and the procedure of Simultaneous Localization and Mapping (SLAM) is demonstrated for a 100 day test approach to the asteroid (16) Psyche, where closed-loop maneuvers are shown to successfully target orbit insertion. A Closed-Loop Linear Covariance (LinCov) analysis approximates the true dispersions and estimation errors, validated against a Monte Carlo study using ideal true measurements. Further tests show the feasibility of incorporating real images into the filter. Image extraction and incorporation is demonstrated along the Psyche approach with Blender simulated data, with additional tests performed on real images of Ceres from the Dawn mission.