Multi-Fidelity Uncertainty Propagation for Objects in Cislunar Space

View Video Presentation: https://doi.org/10.2514/6.2022-1774.vid Cislunar space domain awareness requires accurately quantifying the uncertainty of space objects over extended time intervals. Unlike Earth orbit, measurement cadences are expected to be as large as days to weeks for this regime. Additionally, weak gravitational interactions from the Earth, Sun, and Moon cause highly non-linear dynamical behavior, which in turn produces difficulties in uncertainty propagation between measurement gaps. This work demonstrates the use of a multi-fidelity approach to efficiently yet accurately propagate the orbit state of cislunar space objects with the intended use of rapid uncertainty propagation via particles or sigma points. The state uncertainty is represented as an ensembles of particles and a low-fidelity dynamics model propagates this ensemble forward in time. The method of stochastic collocation then selects a small number of important points from this ensemble, which are then re-propagated from the starting epoch with a high-fidelity model. The re-propagated important points subsequently correct the low-fidelity ensemble. The paper demonstrates the efficacy of this approach for several lunar orbits and a cislunar near-rectilinear halo orbit.