A Novel LiDAR Anchor Constraint Method for Localization in Challenging Scenarios

Positioning system is a key module of autonomous driving. As for LiDAR SLAM system, it faces great challenges in scenarios where there are repetitive and sparse features. Without loop closure or measurements from other sensors, odometry match errors or accumulated errors cannot be corrected. This paper proposes a construction method of LiDAR anchor constraints to improve the robustness of the SLAM system in the above challenging environment. We propose a robust anchor extraction method that adaptively extracts suitable cylindrical anchors in the environment, such as tree trunks, light poles, etc. Skewed tree trunks are detected by feature differences between laser lines. Boundary points on cylinders are removed to avoid misleading. After the appropriate anchors are detected, a factor graph-based anchor constraint construction method is designed. Where direct scans are made to anchor, direct constraints are constructed. While in the position where the anchor is not directly observed, the prior information in the structured road is used to construct a constraint group to improve the overall positioning. When a previously established anchor point is observed at any remote location, the global trajectory can be corrected without waiting for loop closure to occur. We verify the improvement theoretically and experimentally. The result of experiment shows that under the LeGO-LOAM-based factor graph framework, the odometry is improved significantly. The pose correction is realized by long-distance anchor detection when loop closure is not fully realized. Our method can be easily plugged in various factor graph-based systems and has good adaptability and scalability.