InLIOM: Tightly-Coupled Intensity LiDAR Inertial Odometry and Mapping

Hanqi Wang,Huawei Liang,Zhiyuan Li,Xiaokun Zheng,Haitao Xu,Pengfei Zhou,Bin Kong

DOI: https://doi.org/10.1109/tits.2024.3370235

IF: 8.5

2024-01-01

IEEE Transactions on Intelligent Transportation Systems

Abstract:State estimation and mapping are vital prerequisites for autonomous vehicle intelligent navigation. However, maintaining high accuracy in urban environments remains challenging, especially when the satellite signal is unavailable. This paper proposes a novel framework, InLIOM, which tightly couples LiDAR intensity measurements into the system to improve mapping performance in various challenging environments. The proposed framework introduces a stable intensity LiDAR odometry based on scan-to-scan optimization. By extracting features pairwise from intensity information of consecutive frames, this method tackles the instability issue of LiDAR intensity. To ensure the odometry’s robustness, a training-free residual-based dynamic objects filter module is further integrated into the scan-to-scan registration process. The obtained intensity LiDAR odometry solution is incorporated into the factor graph with other multi-sensors relative and absolute measurements, obtaining global optimization estimation. Experiments in indoor and outdoor urban environments show that the proposed framework achieves superior accuracy to state-of-the-art methods. Our approach can robustly adapt to high-dynamic roads, tunnels, underground parking, and large-scale urban scenarios.