New LiDAR-IMU SLAM framework improves docking accuracy for autonomous modular buses

Researchers have developed an enhanced LiDAR-IMU SLAM framework designed to improve localization accuracy and robustness during autonomous modular bus docking, a step that could help make autonomous modular bus systems more reliable, thereby supporting safer, smoother, and more energy efficient next-generation public transport.

Autonomous modular buses have attracted attention as a new model for public transportation because they can dock and undock while in motion, potentially easing traffic congestion and reducing energy use through more flexible operations. But for these systems to work reliably, the vehicles must be able to localize themselves with high precision in both horizontal and vertical directions while also coping with dynamic obstacles and close-range occlusions.

Existing LiDAR-based simultaneous localization and mapping methods can perform well in static environments, but docking scenarios create additional challenges. In particular, the researchers note that conventional methods can suffer from vertical drift and reduced robustness when a nearby vehicle blocks part of the sensor view during docking.

To address these issues, the team proposed an enhanced LiDAR-Inertial Measurement Unit SLAM framework tailored to autonomous modular bus docking. The method combines a two-stage scan-to-map matching strategy with ground constraints to reduce z-axis drift, a factor graph optimization approach that integrates IMU roll and pitch constraints with periodic resetting to limit long-term drift, and a deep learning-based front-vehicle detection and point-cloud filtering mechanism to reduce the effects of occlusion.

In experimental evaluations using real-world single-vehicle and dual-vehicle datasets, the framework reduced Absolute Pose Error and Relative Pose Error compared with existing methods. The results suggest that the approach can improve vertical localization accuracy during docking while remaining more robust in dynamic, obstacle-rich environments.

The study points to a practical pathway for improving autonomous docking performance in modular bus systems, where even relatively small localization errors can affect safety, efficiency, and ride quality. If further validated in broader deployment scenarios, the framework could support more reliable vehicle coordination in intelligent public transportation networks.

Because the reported findings are based on experimental datasets and a specific docking setting, additional work will still be needed to assess performance under a wider range of weather, traffic, and infrastructure conditions. Even so, the results suggest that better sensor fusion and occlusion handling may play an important role in bringing autonomous modular bus concepts closer to real-world operation.

Reference

Author:

Yixu He ^a, Yushu Gao ^b, Yang Liu ^{b c}, Xiaobo Qu ^{b c}

Title of original paper:

LiDAR-IMU SLAM framework in autonomous modular bus docking systems

Article link:

https://www.sciencedirect.com/science/article/pii/S2773153725000933

Journal:

Green Energy and Intelligent Transportation

DOI:

10.1016/j.geits.2025.100343

Affiliations:

^a School of Civil Engineering and Transportation, South China University of Technology, Guangzhou 510640, China

^b School of Vehicle and Mobility, Tsinghua University, Beijing 100084, China

^c State Key Laboratory of Intelligent Green Vehicle and Mobility, Tsinghua University, Beijing 100084, China