While orbital propagators have been investigated extensively over the last fifty years, the consistent propagation of state covariances and more general (non- Gaussian) probability densities has received relatively little attention. The representation of state uncertainty by a Gaussian mixture is well-suited for problems in space situational awareness. Advantages of this approach which are demonstrated in this paper include the potential for long-term propagation in data-starved environments, the capturing of higher-order statistics and more accurate representation of nonlinear dynamical models, the ability to make the filter adaptive using realtime metrics, and parallelizability. Case studies are presented establishing uncertainty consistency and the effectiveness of the proposed adaptive Gaussian sum filter.
展开▼
