A LiDAR-IMU-GNSS fused mapping method for large-scale and high-speed scenarios
Zhaohui Shen,Jibo Wang,Chenglin Pang,Zhengyang Lan,Zheng Fang
DOI: https://doi.org/10.1016/j.measurement.2023.113961
2023-12-04
Measurement: Journal of the International Measurement Confederation
Abstract:In the field of autonomous navigation and surveying, constructing large-scale and globally consistent environmental maps in real-time is a critical problem. In this case, the algorithm faces many challenges, such as high-speed motion, GNSS-denied tunnels and open fields with few features. To solve the aforementioned problems, this paper proposes a mapping method that fuses LiDAR, IMU and GNSS information simultaneously. The proposed method mainly includes two parts: LiDAR-IMU joint optimization and LIO-GNSS joint optimization. In the LiDAR-IMU joint optimization, we realize a tightly-coupled LiDAR inertial odometry (LIO) and propose a new degradation processing mechanism that enables more robust mapping in degraded environments, such as tunnels and open fields. Furthermore, we propose a new constraint method based on ground points to solve the drift problem in the vertical direction. This method maintains a local map based on ground grids, providing more effective constraints for the vertical direction. In the LIO-GNSS joint optimization, we use the factor graph optimization(FGO) algorithm based on sliding window to realize the fusion of LIO and GNSS data, so as to achieve real-time, no drift and global consistency mapping. We conducted experiments in many complex environments with challenges such as high-speed motion, LiDAR degradation, GNSS-denied scenarios and so on. While most open-source algorithms struggled to produce satisfactory mapping results or even failed, our algorithm proved to be robust and achieved the best mapping results.
English Else