Application Research of UWB Indoor Reverse Positioning Based on Robust Kalman Filtering

Abstract

The non-line-of-sight (NLOS) observation errors and outliers generated by the existence of obstacles is a challenge in Ultra-Wideband (UWB) indoor dynamic positioning. To solve this problem, a dynamic robust Kalman filtering model based on indoor reverse positioning is studied. The model uses an Improved Chan-Taylor (I-C-T) algorithm based on height component constrained to obtain positioning observations. For the gross error in the observations, a detection algorithm based on Chi-square increment is introduced. The effect of gross error can be reduced by adjusting the covariance matrix of the robust factor. The experimental results show that the positioning accuracy of the robust Kalman filtering algorithm based on Chi-square increment (CI-RKF) in the x direction is 18.5 cm, which is 23.7%, 26.4% and 38.0% higher than that of conventional robust Kalman filtering (C-RKF) algorithm, Kalman filtering (KF) algorithm and I-C-T algorithm, respectively. The positioning accuracy in the y direction is 11.2 cm, which is 39.2%, 42.8% and 44.6% higher than that of C-RKF algorithm, KF algorithm and I-C-T algorithm, respectively. The results demonstrate the improvement of the reliability and accuracy of indoor positioning, which pave the way for the future dynamic positioning with high accuracy.