Efficient Estimation and Uncertainty Quantification in Space Mission Dynamics

The problem of efficient and accurate orbit estimation of space trajectories is discussed. For highly sensitive low-fuel trajectories designed to exploit the complex nonlinear dynamics of the three-body problem, it is vital to have accurate state estimation during maneuvers and ability to deal with irregular observation update times. For instance, in Halo-orbit insertion and station keeping maneuvers, state estimation errors can propagate quickly. In this paper, we combine an efficient probability propagation method with a homotopy-based posterior computation method. The resulting particle filter is highly accurate even in highly nonlinear regime with intermittent observations, and yet an order of magnitude or more efficient than a generic particle filter implementation.