VBRO: Vehicle Behavior-Guided Reliable Road Surface Constraints for LiDAR Self-Localization in Urban Environments

作者：Xiaofeng Li, Yulong Huang, Shubin Si, Hanxuan Zhang, Ji Yang, Yonggang Zhang · 发表于：IEEE Transactions on Aerospace and Electronic Systems · 年份：2025 · DOI：10.1109/taes.2025.3592629 · 被引用次数：7 · 研究领域：Robotics and Sensor-Based Localization、Autonomous Vehicle Technology and Safety、Remote Sensing and LiDAR Applications

The light detection and ranging (LiDAR) odometry has been widely deployed to realize self-localization of ground vehicles traveling in urban environments. The LiDAR odometry is essentially a dead-reckoning system so that it suffers from serious accumulated errors when external global information is disabled, and the incorrect data association will further exacerbate the error accumulations. To alleviate the problem, a LiDAR self-localization method is proposed based on vehicle behavior-guided reliable road surface constraints. First, the vehicle motion behavior is taken as prior information to refine LiDAR ground observations, which essentially converts the problem of ground estimation from a likelihood estimation to a posteriori estimation, significantly improving ground observation and data association accuracy. Second, the geometric invariant properties and spatiotemporal-scale information of the local covisible grounds are fully considered by the proposed method, and a regionwise spatiotemporal-scale ground constraint with superior environmental adaptability is designed. It allows motion constraints to be enhanced hierarchically by local grounds, significantly inhibiting accumulated errors. Finally, the effectiveness and superiority of the proposed method are validated through a sufficient number of comparisons with the state-of-the-art methods on both public datasets and real-world experiments. The proposed LiDAR-only method achieves more superior localization performanc...