UWB-LiDAR Tightly Coupled Indoor Positioning System for Mobile Robots

This study addresses the challenge of high-precision positioning for mobile robots in complex indoor environments by integrating the concept of Embodied Intelligence (EI). We propose a UWB base station selection strategy based on LiDAR environmental perception, combined with a dynamically adaptive Unscented Kalman Filter (UKF) algorithm. Embodied Intelligence emphasizes the ability of robots to achieve dynamic perception and adaptive decision-making through physical interaction with the environment. Within this framework, our research enhances the adaptability and accuracy of mobile robots in complex environments through multi-sensor fusion. To improve UWB positioning accuracy under Non-Line-of-Sight (NLOS) conditions, the system employs LiDAR to perceive environmental information in real-time, dynamically identify obstacles, select available UWB base stations, and adaptively adjust sensor weights during the data fusion process. Experimental results demonstrate that, compared to using LiDAR or UWB sensors individually, as well as traditional UKF algorithms, the proposed method significantly improves positioning accuracy. Compared to the traditional UKF, the maximum errors in the X and Y directions are reduced by 43.4% and 44.9%, respectively, while the root mean square error (RMS) in the X direction is reduced by 21.3% and by 47.2% in the Y direction. This method effectively addresses UWB measurement errors under NLOS conditions while compensating for the cumulative error of LiDAR, showing significant improvements in both accuracy and robustness. This innovative approach offers an effective solution for high-precision indoor positioning in complex dynamic environments.