Wu, Zhenyu (Nanyang Technological University), Wang, Wei (Nanyang Technological University), Shen, Hongming (Nanyang Technological University), LYU, QIYANG (Nanyang Technological University), Wen, Mingxing (China-Singapore International Joint Research Institute), Peng, Guohao (Nanyang Technological University, Singapore), Wang, Danwei (Nanyang Technological University)

C3M: Collaborative Magnetic-Aware Map-Merging for Intelligent Vehicles in Degraded Environments

Scheduled for presentation during the Regular Session "S43c-Multi-Sensor Fusion and Perception for Robust Autonomous Driving" (FR-LA-T43), Friday, November 21, 2025, 16:40−17:00, Stradbroke

2025 IEEE 28th International Conference on Intelligent Transportation Systems (ITSC), November 18-21, 2025, Gold Coast, Australia

This information is tentative and subject to change. Compiled on October 18, 2025

Abstract

Collaborative mapping is crucial for cooperative unmanned ground vehicles (UGVs) to enhance their understanding of the surrounding environments and to improve the task execution efficiency. However, current collaborative mapping solutions suffer from the perceptual aliasing and geometric data association ambiguities in GNSS-challenged degraded environments (e.g., office/hotel repetitive corridors, long tunnels, industrial warehouses). The ambient magnetic field (MF) has exhibited ubiquity and high distinctiveness at different location regardless of the geometric features, which makes it suitable for positioning and detecting overlapping regions among multiple UGVs' operations. In this paper, a novel Collaborative Magnetic-aware Map-Merging framework, namely C3M, is proposed to cope with the geometrically degraded environments. The key novelties of this work are the MF-based overlapping region delimitation module and the systematic modeling of collaborative magnetic-aware map-merging framework. At single UGV exploring phase, the local magnetic map and point cloud map is generated by fusing information from heterogeneous sensors. At collaborative UGVs rendezvous-triggered phase, local MF sequences are shared and matched to detect overlapping regions, then the transformation matrix is estimated and refined. Eventually, all local maps are merged to obtain the global magnetic map and point cloud map. Evaluations on both high-fidelity simulations and real world experiments show the high quality of the obtained global maps, demonstrating the improved consistency and accuracy of collaborative mapping in challenging degraded environments.