Edge Device-Optimized LiDAR SLAM for Real-Time and Robust Localization in Dynamic Environments

Sai Parimi and Robert Bensch

Abstract:	LiDAR-based Simultaneous Localization and Mapping (SLAM) is a key technology for achieving reliable localization and accurate 3D reconstruction of the environment. In particular, LiDAR SLAM provides robust information when traversing uncharted or GNSS-denied areas, operating reliably under diverse environmental conditions—unlike cameras, which require favorable lighting. This makes LiDAR SLAM central to autonomous navigation, where robots and vehicles demand precise positioning in challenging terrains, and it offers strong benefits when fused with GNSS-based positioning. There has been extensive progress in LiDAR SLAM as well as in algorithms for dynamic object removal. However, relatively little attention has been given to integrating these two aspects into a unified framework capable of real-time operation. In this work, we address this gap by presenting a real-time LiDAR SLAM pipeline that seamlessly incorporates dynamic object removal to achieve robust performance in highly dynamic environments. Our pipeline tightly couples LiDAR pointclouds with Inertial Measurement Unit (IMU) data in the frontend to produce accurate odometry estimates. The estimated trajectory, along with the associated pointclouds, is then optimized in a factor graph-based backend to reduce long-term drift and improve global consistency. To further strengthen performance, we incorporate dynamic object removal, which improves odometry estimation by emphasizing stable environmental features while also producing cleaner global maps. These refined maps enable more reliable global registration and enhance the subsequent re-localization process. Robust re-localization, in turn, allows drift-free traversal within pre-built maps. Finally, we demonstrate how fusing our improved LiDAR-Inertial Odometry with Real-Time Kinematic (RTK) positioning further enhances localization accuracy, particularly in GNSS-denied environments.
Published in:	Proceedings of the 38th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2025) September 8 - 12, 2025 Hilton Baltimore Inner Harbor Baltimore, Maryland
Pages:	2021 - 2030
Cite this article:	Parimi, Sai, Bensch, Robert, "Edge Device-Optimized LiDAR SLAM for Real-Time and Robust Localization in Dynamic Environments," Proceedings of the 38th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2025), Baltimore, Maryland, September 2025, pp. 2021-2030. https://doi.org/10.33012/2025.20450
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In