LL-Localizer: A Life-Long Localization System based on Dynamic i-Octree

TL;DR

This work tackles robust life-long localization under multi-session and evolving environments by proposing LL-Localizer, a system that relies on a prior map and incremental voxel updates. A three-layer map (prior $\mathcal{M}^p$, temporary $\mathcal{M}^t$, and static $\mathcal{M}^s$) is managed by a Dynamic $i$-Octree to enable efficient local-map loading and updates, while IMU data supports high-frequency pose prediction. The method introduces a three-stage scan processing pipeline with a nuanced local-map loading strategy, robust point-cloud registration, and selective map updates to handle unmapped areas and environmental changes. Experimental results on NCLT, M2DGR, and BG datasets show localization accuracy comparable to state-of-the-art LO/LIO systems, with strong resilience to environmental changes and unknown regions, and real-time performance enabled by the Dynamic $i$-Octree.

Abstract

This paper proposes an incremental voxel-based life-long localization method, LL-Localizer, which enables robots to localize robustly and accurately in multi-session mode using prior maps. Meanwhile, considering that it is difficult to be aware of changes in the environment in the prior map and robots may traverse between mapped and unmapped areas during actual operation, we will update the map when needed according to the established strategies through incremental voxel map. Besides, to ensure high performance in real-time and facilitate our map management, we utilize Dynamic i-Octree, an efficient organization of 3D points based on Dynamic Octree to load local map and update the map during the robot's operation. The experiments show that our system can perform stable and accurate localization comparable to state-of-the-art LIO systems. And even if the environment in the prior map changes or the robots traverse between mapped and unmapped areas, our system can still maintain robust and accurate localization without any distinction. Our demo can be found on Blibili (https://www.bilibili.com/video/BV1faZHYCEkZ) and youtube (https://youtu.be/UWn7RCb9kA8) and the program will be available at https://github.com/M-Evanovic/LL-Localizer.

Figures (6)

• Figure 1: Motivation for our incremental voxel-based life-long localization approach. The white point cloud denotes the prior map loaded at startup, while the green point cloud highlights incremental updates to accommodate new areas or environmental changes (e.g., traffic, seasons). This functionality is crucial for long-duration service robots.
• Figure 2: Diagram of Dynamic i-Octree.
• Figure 3: System overview of LL-Localizer.
• Figure 4: The trajectory results of different systems in bg_2 (left) and bg_3 (right).
• Figure 5: The error of different systems in bg_2 (left) and bg_3 (right).