.png){kind=icon}
.png){kind=icon}
Return
LEMON-Mapping: Loop-Enhanced Large-Scale Multi-Session Point Cloud Merging and Optimization for Globally Consistent Mapping
.png){kind=icon}
PRE
AI
DOI：10.1109/tase.2026.3709653.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Multi-robot collaboration is critical but challenging for building globally consistent maps. Traditional multi-robot pose graph optimization (PGO) methods ensure basic global consistency but ignore the geometric structure of the map and only use loop closures as constraints between pose nodes, which leads to divergence and blurring in overlapping regions. To address this, we propose LEMON-Mapping, a loop-enhanced framework for large-scale, multi-session point cloud fusion and optimization. We re-examine the role of loops in multi-robot mapping and present three key innovations. First, we develop a robust loop processing mechanism with outlier rejection and a recall strategy to recover valid loops. Second, we introduce spatial bundle adjustment to reduce divergence and eliminate blurring in overlapping areas. Third, we design a PGO-based optimization that integrates refined bundle adjustment constraints to propagate local accuracy globally. Experiments on public and self-collected datasets demonstrate that LEMON-Mapping achieves superior accuracy, consistency, and scalability over traditional approaches in large-scale multi-robot scenarios. Note to Practitioners—In this paper, we address inaccurate and inconsistent large-scale multi-robot map fusion, where conventional multi-robot SLAM ignores geometric structures, causing divergence and blurring in overlapping regions. In real-world exploration, multi-robot mapping is designed to extend coverage, but inconsistent and inaccurate map fusion degrades map quality and affects downstream applications like re-localization, navigation, and robotic operation, potentially leading to navigation failures and unsafe behaviors. LEMON-Mapping integrates robust loop processing, spatial bundle adjustment, and pose graph optimization to reconstruct accurate and globally consistent multi-robot maps, providing reliable support for multi-robot collaboration, localization, and autonomous navigation.
Keywords:
Localization and mapping
multi-robot SLAM
swarms
Journal
IF:
6.4
Papers:
4.9K
Citations:
1.6W