3D Foundation Model-Based Loop Closing for Decentralized Collaborative SLAM

TL;DR

This work tackles decentralized collaborative SLAM under large viewpoint variations by leveraging 3D foundation models to infer inter-robot relative poses from monocular image pairs without sharing full maps. It integrates MASt3R into a modular C-SLAM pipeline with CosPlace-based place recognition, a confidence-weighted MASt3R registration, and scale-aware pose-graph optimization to fuse multi-robot maps. The authors propose three factor-graph formulations—base, independent scales, and smoothed scales—to explicitly handle loop-closure scale and enable distributed optimization with a dynamically elected solver. Empirical results on multi-robot sequences show improved localization accuracy and substantial gains in computational and memory efficiency compared with state-of-the-art decentralized baselines, validating the approach's scalability for large-scale deployments. The work highlights practical potential for real-time, bandwidth-efficient multi-robot mapping in unknown environments, while noting dependencies on metric-scale odometry and generalization considerations for foundation models.

Abstract

Decentralized Collaborative Simultaneous Localization And Mapping (C-SLAM) techniques often struggle to identify map overlaps due to significant viewpoint variations among robots. Motivated by recent advancements in 3D foundation models, which can register images despite large viewpoint differences, we propose a robust loop closing approach that leverages these models to establish inter-robot measurements. In contrast to resource-intensive methods requiring full 3D reconstruction within a centralized map, our approach integrates foundation models into existing SLAM pipelines, yielding scalable and robust multi-robot mapping. Our contributions include: (1) integrating 3D foundation models to reliably estimate relative poses from monocular image pairs within decentralized C-SLAM; (2) introducing robust outlier mitigation techniques critical to the use of these relative poses; and (3) developing specialized pose graph optimization formulations that efficiently resolve scale ambiguities. We evaluate our method against state-of-the-art approaches, demonstrating improvements in localization and mapping accuracy, alongside significant gains in computational and memory efficiency. These results highlight the potential of our approach for deployment in large-scale multi-robot scenarios.