Chain-SLAM: Globally Consistent Backend for Multi-Session LiDAR SLAM via Chained Loop Closure

Zhiheng Li, Xinhao Liu, Juexiao Zhang, Yongqing Liang, Chen Feng
New York University
IROS 2026
arXiv Code Data (ROS bags)
Chained loop closures across mapping sessions

Direct and chained loop closures between a current session i and loaded sessions j, k. Each successful closure from the current session triggers multiple chained closures to the loaded sessions, enhancing multi-session consistency.

Abstract

Maintaining consistency over long spatial and temporal horizons remains a fundamental challenge in large-scale LiDAR SLAM, particularly when integrating maps collected across multiple sessions. We present Chain-SLAM, a LiDAR SLAM backend enabling online multi-session map alignment and reuse with global consistency at large scale. We implement a chained loop closure mechanism that efficiently propagates geometric constraints across inter-session keyframes through an adjacency graph, enabling robust long-horizon consistency triggered by reliable short-horizon loop closures. The system initializes inter-session alignment with GNSS-proximity place recognition, then performs on-the-fly loop closure detection and joint optimization of loaded maps and newly acquired trajectories within a unified factor graph, maintaining both inter- and intra-session geometric consistency without dynamic object removal, and cross-platform robustness with minimal hyperparameter tuning. Experimental results show improved trajectory accuracy and robust multi-session integration on large-scale datasets. We release our source code to support reproducible research in large-scale multi-session LiDAR SLAM.

Chained Loop Closure

Chained closure detection and verification pipeline
Starting from a new direct loop closure, a BFS over the pose graph gathers connected keyframes, ICP verifies each candidate chained closure (accepting those with residual eij ≤ τ), and the verified constraints are added as new edges to the adjacency graph.

Results

Trajectory evaluation across sessions
Estimated trajectories across multiple sessions (top vs. bottom rows) overlaid per session, with the per-session alignment error against ground truth shown on the right. Evaluated on the multi-session MARS and NCLT datasets.

BibTeX

@inproceedings{li2026chainslam,
  title     = {Chain-SLAM: Globally Consistent Backend for Multi-Session LiDAR SLAM via Chained Loop Closure},
  author    = {Li, Zhiheng and Liu, Xinhao and Zhang, Juexiao and Liang, Yongqing and Feng, Chen},
  booktitle = {IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
  year      = {2026}
}