LRSLAM: Low-Rank Representation of Signed Distance Fields in Dense Visual SLAM System

2025-06-12

European Conference on Computer Vision

AI Summary

Dense visual SLAM faces fundamental bottlenecks in real-time performance, robustness, and scalability to large-scale scenes—stemming from high computational overhead, excessive memory consumption, and slow optimization convergence. To address these challenges, this paper proposes an efficient Signed Distance Field (SDF) representation based on low-rank tensor decomposition, introducing for the first time a joint Six-axis and CANDECOMP/PARAFAC (CP) decomposition framework for implicit SDF modeling. This approach substantially curbs parameter growth and accelerates optimization convergence. Integrated within an RGB-D dense SLAM pipeline leveraging implicit neural field optimization, our method achieves a 42% reduction in model parameters and a 3.1× speedup in processing time across multiple indoor benchmarks, while simultaneously improving reconstruction completeness and pose estimation accuracy over state-of-the-art methods. The core contribution lies in a low-rank tensor-driven lightweight SDF representation that effectively balances efficiency and geometric fidelity.

Technology Category

• Computer Vision: Representation Learning for Vision
• Machine Learning: Matrix & Tensor Methods
• Intelligent Robots: State Estimation

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Simultaneous Localization and Mapping (SLAM) has been crucial across various domains, including autonomous driving, mobile robotics, and mixed reality. Dense visual SLAM, leveraging RGB-D camera systems, offers advantages but faces challenges in achieving real-time performance, robustness, and scalability for large-scale scenes. Recent approaches utilizing neural implicit scene representations show promise but suffer from high computational costs and memory requirements. ESLAM introduced a plane-based tensor decomposition but still struggled with memory growth. Addressing these challenges, we propose a more efficient visual SLAM model, called LRSLAM, utilizing low-rank tensor decomposition methods. Our approach, leveraging the Six-axis and CP decompositions, achieves better convergence rates, memory efficiency, and reconstruction/localization quality than existing state-of-the-art approaches. Evaluation across diverse indoor RGB-D datasets demonstrates LRSLAM's superior performance in terms of parameter efficiency, processing time, and accuracy, retaining reconstruction and localization quality. Our code will be publicly available upon publication.

Problem

• Improving real-time performance in dense visual SLAM
• Reducing memory usage for large-scale scene mapping
• Enhancing reconstruction and localization accuracy efficiently

Innovation

• Low-rank tensor decomposition for SLAM
• Six-axis and CP decompositions for efficiency
• Improved memory efficiency and accuracy