Observability Analysis and Performance Evaluation for a Graph-Based GNSS–Visual–Inertial Odometry on Matrix Lie Groups

Shu-Hua Tsao

Peer Reviewed

Abstract:	This paper presents a graph optimization-based global navigation satellite system (GNSS)–visual–inertial odometry (GVIO) for robotic platforms by modeling the state space as the recently proposed matrix Lie groups. The combination of visual–inertial navigation with raw GNSS observations opens up the possibilities of global localization in geocentric coordinates without drift caused by yaw and translation in the global frame, which are the four unobservable directions. Particularly, the proposed GVIO is extended from the existing graph-based framework with a novel IMU preintegration framework. Analytically, we showed that the states are fully observable with the aid of the raw GNSS observations and the degenerate case with less than four satellites tracked. Furthermore, the analysis and the proposed estimator are extensively validated through publicly available and self-collected real-world datasets.
Published in:	Proceedings of the 36th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2023) September 11 - 15, 2023 Hyatt Regency Denver Denver, Colorado
Pages:	2033 - 2047
Cite this article:	Tsao, Shu-Hua, "Observability Analysis and Performance Evaluation for a Graph-Based GNSS–Visual–Inertial Odometry on Matrix Lie Groups," Proceedings of the 36th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2023), Denver, Colorado, September 2023, pp. 2033-2047. https://doi.org/10.33012/2023.19401
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In