Neural Graph Mapping for Dense SLAM with Efficient Loop Closure

2024-05-06

arXiv.org

AI Summary

Existing neural radiance field (NeRF)-based SLAM approaches rely on a monolithic volumetric representation, suffering from inefficient loop-closure integration and poor scalability. To address this, we propose Neural Graph Mapping—a novel framework that anchors lightweight MLP-based implicit fields to nodes of a sparse pose graph constructed by visual SLAM, enabling joint optimization of neural fields and pose graph. This paradigm supports loop-closure-driven local re-integration and incremental field updates, overcoming fundamental limitations of global neural fields in loop consistency and large-scale mapping. Evaluated on building-scale scenes with multiple loops, our method achieves superior reconstruction accuracy, real-time performance, and global consistency compared to state-of-the-art methods. It significantly enhances scalability and practicality for large-scale dense 3D reconstruction.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Intelligent Robots: Localization, Mapping, and Navigation
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Existing neural field-based SLAM methods typically employ a single monolithic field as their scene representation. This prevents efficient incorporation of loop closure constraints and limits scalability. To address these shortcomings, we propose a neural mapping framework which anchors lightweight neural fields to the pose graph of a sparse visual SLAM system. Our approach shows the ability to integrate large-scale loop closures, while limiting necessary reintegration. Furthermore, we verify the scalability of our approach by demonstrating successful building-scale mapping taking multiple loop closures into account during the optimization, and show that our method outperforms existing state-of-the-art approaches on large scenes in terms of quality and runtime. Our code is available at https://kth-rpl.github.io/neural_graph_mapping/.

Problem

• Monolithic neural fields hinder loop closure integration in SLAM
• Lack of scalable dense mapping with efficient loop closure handling
• Existing methods struggle with large-scale scene quality and runtime

Innovation

• Lightweight neural fields for scene representation
• Dynamic anchoring to sparse SLAM pose graph
• Efficient large-scale loop closure integration