Covariance intersection algorithm for distributed spacecraft state estimation

The use of a fleet of highly synchronized spacecraft to synthesize a space-based optical interferometer, as envisioned by the TPF mission of NASA Origins program, will allow for baselines large enough to achieve milli-arcsecond resolutions. We explore the use of the covariance intersection (CI) algorithm for state estimation of distributed spacecraft. The CI algorithm provides a mechanism for fusing two or more random variables with unknown cross-covariance such that the computed covariance of the new estimate is consistent. We consider the case of five spacecraft with a decentralized estimation scheme where each spacecraft shares its own estimates with the rest of the fleet using a ring-type communication architecture. This simplified scenario of the TPF mission is used to highlight the problems that can originate if the cross-covariances are completely ignored. The result shows that the CI filter performance instead is not affected and provides good estimates