A Novel UWB/IMU/Odometer-Based Robot Localization System in LOS/NLOS Mixed Environments

The accuracy of existing UWB range-based indoor localization methods is generally degraded due to the non-line-of-sight (NLOS) situations where a serve bias in UWB range measurements is unavoidable. In this article, we first propose a two-stage NLOS detection method to detect line-of-sight (LOS)-measured distances in mixed LOS/NLOS indoor environments. Then, a high-accuracy UWB/IMU/Odometer integrated localization system is presented using an adaptive multi-algorithm localization framework based on the number of detected LOS-measured distances. Specifically, under conditions of one or two LOS-measured distances, an improved adaptive EKF positioning algorithm (IAEKF) is proposed. Compared with the traditional extended Kalman filter (EKF)-based fusion scheme, the weight function of innovation is exploited to adaptively estimate the measurement noise covariance matrices and further reduce the influence of the changing measurement noise in NLOS conditions. For three or more LOS-measured ranges, a novel tightly-coupled fusion factor graph framework is developed. To further improve the performance of seamless positioning in transaction areas, a strong constraint of trajectory smoothness is designed and added to the factor graph framework by using the weight value of IMU/odometer measurements. The experimental results show that the proposed localization system achieves an average localization error of 0.227 m, which surmounts UWB range-based and integrated methods in LOS/NLOS mixed environments.