Vision-based, terrain-aided navigation with decentralized fusion and finite set statistics

James S. McCabe and Kyle J. DeMars

Peer Reviewed

Abstract: Terrain-related information, in the form of features extracted from images, presents a rich data source that can be harvested to facilitate drastic improvements in navigation when conventional data sources, such as GPS, are not available. Conventional implementation of such data types requires image correlation techniques that interrupt streamlined transmission of statistics through a navigation filter, oftentimes leading to time-wise correlations that are erroneously ignored. This paper proposes leveraging finite set statistics to recast the terrain feature data into a simultaneous localization and mapping problem. Decentralized data fusion is employed to augment a standard extended Kalman filter-based navigation with the terrain data. Theoretical results are supported with a simulated descent to landing navigation scenario that demonstrates the improvements offered by augmenting standard navigation with terrain aiding.
Published in: NAVIGATION, Journal of the Institute of Navigation, Volume 66, Number 3
Pages: 537 - 557
Cite this article: McCabe, James S., DeMars, Kyle J., "Vision-based, terrain-aided navigation with decentralized fusion and finite set statistics", NAVIGATION, Journal of The Institute of Navigation, Vol. 66, No. 3, Fall 2019, pp. 537-557.
https://doi.org/10.1002/navi.320
Full Paper: ION Members: Free Download
Non-Members: Free Download
Sign In