Multi-Submap Implicit Neural SLAM with Local-to-Global Loop Closure for Large-Scale Scene Reconstruction

2026-08-10

AI Summary

This work addresses the failure of NeRF-based SLAM in large-scale, complex environments caused by catastrophic forgetting and trajectory drift. To overcome these challenges, we propose a multi-submap implicit neural SLAM system that employs a progressive mapping strategy with dynamically allocated neural submaps. Our approach integrates optical flow tracking, a dual-layer local–global loop closure detection mechanism, foundation-model-driven global descriptor extraction, and online knowledge distillation across submap boundaries to jointly preserve geometric and photometric consistency. Extensive experiments on both public and custom large-scale indoor–outdoor datasets demonstrate that our system significantly outperforms existing neural SLAM methods in reconstruction fidelity and pose estimation robustness. Furthermore, real-world deployment is validated through a customized handheld platform, confirming its practical applicability.

Abstract

Neural Radiance Fields (NeRF)-based SLAM has demonstrated impressive results in small-scale scene reconstruction, yet scaling these methods to extensive, complex environments remains challenging due to catastrophic forgetting and accumulated trajectory drift. This paper presents a robust, large-scale neural SLAM system featuring a multi-submap architecture and a dual-tier loop closure mechanism. Specifically, we propose a progressive mapping strategy that dynamically allocates neural submaps to maintain high-fidelity representations without memory explosion. For robust pose estimation, an optical-flow-based tracking module is integrated to handle aggressive motions. To address global consistency, we introduce a local-to-global loop closure framework leveraging the foundation model for high-performance global descriptor extraction, significantly enhancing relocalization accuracy under varying viewpoints. Furthermore, an inter-submap online distillation algorithm is designed during back-end optimization to enforce geometric and appearance consistency across overlapping submap boundaries. To validate the system, we developed a customized handheld mechatronic platform and conducted extensive evaluations on both public benchmarks and our large-scale indoor-outdoor datasets. Experimental results, including direct deployment on an onboard computing unit, demonstrate that our approach outperforms state-of-the-art neural SLAM methods in reconstruction quality and localization robustness, providing a scalable solution for real-world robotic perception and digital twinning. We will release the code publicly on \href{https://github.com/dtc111111/MSN-SLAM}{https://github.com/dtc111111/MSN-SLAM} .

Problem

• Neural SLAM
• large-scale reconstruction
• catastrophic forgetting
• trajectory drift
• global consistency

Innovation

• Neural SLAM
• Multi-submap Architecture
• Local-to-Global Loop Closure
• Online Distillation
• Foundation Model