Performance Evaluation of Left-/Right-invariant EKF According to Appropriate Model Selection

Jaehyuck Cha, Jeong Ho Hwang, Chan Gook Park

Abstract:	This paper deals with invariant extended Kalman filter (IEKF) which has a trajectory-independent property by defining the error state using the Lie group, unlike the conventional Kalman filter. Inertial navigation system (INS) is a widely used navigation solution providing 6 degrees of freedom motion. Since the performance of the INS tends to deteriorate with time, it requires another aiding sensor. In the fusion of the INS and such aiding sensor, the Kalman filter or its variants are mainly adopted. Specifically, it is well-known that an IEKF is effective when the initial attitude estimation error is large. There are two types of IEKFs, left IEKF (LIEKF) and right IEKF (RIEKF). It is known to be advantageous to adopt the LIEKF for the left-invariant measurements and right-IEKF for the right-invariant measurements. However, there have been few studies that have analyzed in detail the degradation of performance when it is used with the opposite IEKF. Therefore, this paper evaluates the performance of left-/right-IEKF according to the invariancy of a sensor measurement.
Published in:	Proceedings of the 35th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2022) September 19 - 23, 2022 Hyatt Regency Denver Denver, Colorado
Pages:	1687 - 1693
Cite this article:	Cha, Jaehyuck, Hwang, Jeong Ho, Park, Chan Gook, "Performance Evaluation of Left-/Right-invariant EKF According to Appropriate Model Selection," Proceedings of the 35th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2022), Denver, Colorado, September 2022, pp. 1687-1693. https://doi.org/10.33012/2022.18524
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In