Singular value decomposition‐based iterative robust cubature Kalman filtering and its application for integrated global positioning system/strapdown inertial navigation system navigation

Abstract

The issue of non-linear robust state estimation in the integration of a strapdown inertial navigation system and global positioning system is addressed in this study. Based on the cubature Kalman filtering frame, a non-linear robust filter called a robust cubature Kalman filter (RCKF) was introduced to address the outliers and the inaccurate model. It has been found that the determination of an optimal restriction parameter is crucial for maintaining the robustness and accuracy of the non-linear robust filter. Unfortunately, the value of this restriction parameter is always determined by experience. In this study, an iterative strategy was proposed to adaptively attain the optimal restriction parameter without much previous experience. To improve the computational stability of the iterative non-linear robust filter, a singular value decomposition strategy was adopted simultaneously. Two case studies indicate that the iterative RCKF can achieve greater robustness and accuracy using the methodology discussed in this study. K E Y W O R D S cubature Kalman filter, integrated navigation system, non-linear filter, singular value decompositionThis is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.

Double‐state chi‐square test based sparse grid quadrature filtering algorithm and its application in integrated navigation

“…In order to meet the high accuracy requirement of nonlinear system, the differential free filtering (DFF) has become a research priority [9]. The unscented Kalman filter (UKF) proposed by Julier [10] and the cubature Kalman filter (CKF) proposed by Arasaratnam [11] are representative of the DFF. The UKF and its improved filter approximation of the posterior distribution can be accurate to the third order of the Taylor expansion.…”