Sequential estimation methods for small body optical navigation
MetadataShow full item record
As humans explore further into the solar system, small bodies such as asteroids and comets serve as critical stepping-stone destinations. Highly accurate navigation about these small bodies is critical for any future missions, and as a result is listed prominently among NASA's future goals in the NASA Office of Chief Technologist Roadmap. Due to the long communication light-time delays with the Earth, advances in small body navigation may enable missions currently not feasible, as well as significantly reduce dependence on ground resources. Increased operational agility will enable rapid decisions and opportunistic science measurements not possible in previous missions to small bodies. To assist NASA in accomplishing future small body navigation goals, several important advances are made. First, the effectiveness of modern orbit estimation techniques is investigated, with the higher order Additive Divided-Difference sigma point Filter (ADF) implemented and used along with the standard Extended Kalman Filter (EKF) to estimate the spacecraft state from optical small body surface landmark measurements. The ADF performs consistently better than the EKF in the simulations performed, with increasing improvement for higher levels of initial state error and longer intervals between photos of the surface. Second, a new method is created to improve onboard navigation filter performance in diverse and rapidly changing dynamical environments. The approach is to precompute a process noise profile along a reference trajectory using consider covariance analysis tools and filters. When used in an onboard navigation filter, the precomputed process noise allows the filter to account for time- and state-dependent perturbations in the dynamics. The new method also obviates the need for most or all traditional manual tuning of the filter, and provides significantly improved representation of the state uncertainty. Finally, a Simultaneous Localization And Mapping (SLAM) algorithm is employed to estimate the spin state of a tumbling small body (which are expected to be a significant percentage of the small bodies in the solar system), as well as the spacecraft state and surface landmark locations. For the small body characterization phase of the Rosetta mission, the state estimates converge successfully for large initial state errors. The SLAM algorithm remains effective for a range of small body spin states and masses that correspond to expected tumbling small bodies throughout the solar system. The SLAM algorithm is successfully applied to high fidelity independently simulated imagery of a tumbling small body generated by the European Space Agency, and a method for initializing the small body landmark locations is provided.