INS/UWB fusion localization algorithm in indoor environment based on variational Bayesian and error compensation

Abstract

Localization technology is crucial for indoor robot navigation. However, because of the intricacies inherent in the indoor setting, the signal transmission is vulnerable to the interference of obstacles, which leads to the decline of positioning accuracy. Ultra-Wideband (UWB) has the characteristics of channel insensitivity and high localization accuracy. Inertial navigation system (INS) functions independently as a navigation system, and its positioning results will not be affected due to non-line-of-sight (NLOS) interference. When using UWB to locate the mobile node, the Variational Bayesian Gaussian Mixture Model (VBGMM) clustering algorithm based on Gaussian Mixture Model (GMM) is applied to lessen the influence of NLOS propagation. This paper proposed a loose coupling of the INS and UWB, which combines the advantages of the two subsystems and improves the performance of the positioning system. On the basis of INS autonomous positioning, the maximum entropy fuzzy generalized probability data association filter (MEF-GPDAF) is used to modify the INS positioning results, and then the virtual inertia points are further built to compensate the error of the corrected coordinates. Finally, Unscented Kalman Filter is applied to the compensated coordinates for enhanced positioning. Simulation indicates that the proposed approach in this paper exhibits superior location accuracy. The real experimental results show that the proposed algorithm achieves an average improvement of 61.12% in positioning accuracy.