A High-Accuracy Indoor Localization System and Applications Based on Tightly Coupled UWB/INS/Floor Map Integration

Changqiang Wang,Aigong Xu,Jian Kuang,Xin Sui,Yushi Hao,Xiaoji Niu

DOI: https://doi.org/10.1109/jsen.2021.3082579

IF: 4.3

2021-01-01

IEEE Sensors Journal

Abstract:High-accuracy localization systems for non-line-of-sight (NLOS) environments have been a popular focus of research on indoor positioning. In particular, the use of ultra-wideband (UWB) technology for ranging is emerging as an effective method of indoor localization due to its high accuracy. To improve the overall performance of UWB systems, combining UWB technology with an inertial navigation system (INS) is a commonly used indoor positioning approach. However, UWB measurements are affected by outliers under the NLOS propagation of radio signals. In addition, INS tend to suffer from error accumulation without constraints imposed by observations from other sensors. To address these problems, an adaptive robust extended Kalman filter algorithm based on tightly coupled UWB/INS/floor map integration is proposed, with which the noise in the UWB observations can be adaptively adjusted. The integrated system performs NLOS identification based on a map line segment matching algorithm; then, abnormal observations are abandoned, or their weights are reduced to prevent the error accumulation of the INS from being incorrectly constrained. Finally, an evaluation based on experimental data shows the significant enhancement achieved with the proposed algorithm based on the integration of UWB technology, an INS and a floor map, which can achieve a 2-D positioning accuracy that corresponds to a root mean square error of 0.27 m. Thus, substantial improvements can be achieved for long-term operation in NLOS environments using the proposed integration scheme.