A Tightly Coupled Positioning Method of Ranging Signal and IMU Based on NLOS Recognition

Ke Han,Bingxun Liu,Zhongliang Deng

DOI: https://doi.org/10.1109/ipin54987.2022.9918144

2022-01-01

Abstract:Due to the complex indoor environment, the human body and infrastructure will cause the non-line-of-sight (NLOS) error, resulting in inaccurate ranging information. To effectively use the inertial measurement unit (IMU) to reduce the NLOS error, a tightly coupled localization method based on NLOS signal adaptive recognition is proposed in this paper. Firstly, the method predicts the distance from the base station to terminals using the tight coupling of the IMU and the ranging signal. Secondly, we propose an NLOS recognition method based on the residuals between the observed and predicted values of distance. In the form of a sliding window, the residuals are clustered using a Gaussian mixture model (GMM), and the clustering results are optimized by introducing the concept of time series. The NLOS recognition is made according to the obtained residual distribution model. Then, the error state Kalman filter (ESKF) is performed on the residuals identified as line-of-sight (LOS) to correct the position information of the terminal. Simulation results show that this recognition method has higher sensitivity and accuracy. We also use ultra-wideband (UWB) to test. The results show that in the mixed environment of LOS and NLOS, this method can identify Los signals and use them for positioning, and the accuracy is 47% higher than traditional tight coupling.