Accuracy analysis of sigma-point Kalman filters
Wei David Fan, Yong Li · 2009
Sigma-point Kalman filters are new filters with high precision aimed at nonlinear system. Within the framework of linear minimum variance recursive algorithm, the accuracy of state estimation using the sigma-point Kalman filters mainly depends on the strategies of choosing sigma-points. In this paper, theorems are presented to determine the relationship between the sigma-point Kalman filters' estimate accuracy about the means and variances and the strategies of choosing sigma-points. Then, some deductions about the accuracy of unscented Kalman filter (UKF), divided difference filter (DDF) and Gaussian-Hermite filter (GHF) are presented. The accuracy analysis of state estimation via the sigma-point Kalman filters can benefit from these theorems and deductions.