Research on Multi-Sensor Fusion Localization Methods for Outdoor Inspection Robots

This paper presents a multi-sensor fusion localization algorithm based on IMU/LiDAR/RTK to address the limitations of individual sensors in dynamic and obstructed environments. By incorporating RTK's high-precision position data into the IMU-based state estimation framework, the algorithm compensates for drift errors. Additionally, it undistorts observations by fusing RTK and LiDAR data. The paper employs a custom ikd Tree search method to enhance the efficiency of LiDAR pointcloud matching, thereby ensuring real-time performance. The algorithm uses an Iterative Error State Kalman Filter (IESKF) to minimize residuals, and it is validated through simulations and real-world tests. The results demonstrate improved positioning accuracy and robustness, especially when RTK data reliability is high, thereby significantly reducing absolute position errors.