SP-SLAM: Neural Real-Time Dense SLAM With Scene Priors

TL;DR

SP-SLAM is a real-time neural RGB-D SLAM system that leverages scene priors and efficient encoding to improve reconstruction quality and tracking accuracy without sacrificing speed.

Contribution

It introduces a novel neural SLAM framework that uses sparse voxel scene priors and tri-plane appearance encoding for fast, high-quality dense mapping.

Findings

Achieves superior accuracy on benchmark datasets.

Runs significantly faster than existing neural SLAM methods.

Provides high-fidelity surface reconstruction with efficient memory use.

Abstract

Neural implicit representations have recently shown promising progress in dense Simultaneous Localization And Mapping (SLAM). However, existing works have shortcomings in terms of reconstruction quality and real-time performance, mainly due to inflexible scene representation strategy without leveraging any prior information. In this paper, we introduce SP-SLAM, a novel neural RGB-D SLAM system that performs tracking and mapping in real-time. SP-SLAM computes depth images and establishes sparse voxel-encoded scene priors near the surfaces to achieve rapid convergence of the model. Subsequently, the encoding voxels computed from single-frame depth image are fused into a global volume, which facilitates high-fidelity surface reconstruction. Simultaneously, we employ tri-planes to store scene appearance information, striking a balance between achieving high-quality geometric texture mapping and minimizing memory consumption. Furthermore, in SP-SLAM, we introduce an effective optimization strategy for mapping, allowing the system to continuously optimize the poses of all historical input frames during runtime without increasing computational overhead. We conduct extensive evaluations on five benchmark datasets (Replica, ScanNet, TUM RGB-D, Synthetic RGB-D, 7-Scenes). The results demonstrate that, compared to existing methods, we achieve superior tracking accuracy and reconstruction quality, while running at a significantly faster speed.