INS-GNSS Navigation for Large Attitude Uncertainties with the Matrix Fisher-Gaussian Distribution

In this paper, we present a new recursive Bayesian filter for loosely coupled INS-GNSS navigation to handle large attitude uncertainty. The filter replaces the Gaussian distribution assumed in Kalman filters with the matrix Fisher-Gaussian (MFG) distribution, which is defined intrinsically on the pr...

Full description

Saved in:
Bibliographic Details
Main Authors: Wang, Weixin, Lee, Taeyoung
Format: Conference Proceeding
Language:English
Subjects:
Bayes methods
Kalman filters
Manifolds
Navigation
Pedestrians
Position measurement
Uncertainty
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this paper, we present a new recursive Bayesian filter for loosely coupled INS-GNSS navigation to handle large attitude uncertainty. The filter replaces the Gaussian distribution assumed in Kalman filters with the matrix Fisher-Gaussian (MFG) distribution, which is defined intrinsically on the product manifold of the three dimensional special orthogonal group and the Euclidean space of an arbitrary dimension. The MFG models large attitude uncertainty accurately, which can be frequently encountered in robot and pedestrian localization, for example when the robot or person enters a building where the heading direction is unobservable. It is validated by simulation studies illustrating that the proposed filter has a substantially faster convergence rate, when compared with the extended Kalman filter.
ISSN:2378-5861
DOI:10.23919/ACC55779.2023.10156004