CG-SLAM: Efficient Dense RGB-D SLAM in a Consistent Uncertainty-aware 3D Gaussian Field

TL;DR

This paper presents an efficient dense RGB-D SLAM system, i.e., CG-SLAM, based on a novel uncertainty-aware 3D Gaussian field with high consistency and geometric stability, based on an in-depth analysis of Gaussian Splatting.

Abstract

Recently neural radiance fields (NeRF) have been widely exploited as 3D representations for dense simultaneous localization and mapping (SLAM). Despite their notable successes in surface modeling and novel view synthesis, existing NeRF-based methods are hindered by their computationally intensive and time-consuming volume rendering pipeline. This paper presents an efficient dense RGB-D SLAM system, i.e., CG-SLAM, based on a novel uncertainty-aware 3D Gaussian field with high consistency and geometric stability. Through an in-depth analysis of Gaussian Splatting, we propose several techniques to construct a consistent and stable 3D Gaussian field suitable for tracking and mapping. Additionally, a novel depth uncertainty model is proposed to ensure the selection of valuable Gaussian primitives during optimization, thereby improving tracking efficiency and accuracy. Experiments on various datasets demonstrate that CG-SLAM achieves superior tracking and mapping performance with a notable tracking speed of up to 15 Hz. We will make our source code publicly available. Project page: https://zju3dv.github.io/cg-slam.

Figures (5)

• Figure 1: CG-SLAM, which adopts a well-designed 3D Gaussian field, can simultaneously achieve state-of-the-art performance in localization, reconstruction and rendering. Benefiting from 3D Gaussian representation and a new GPU-accelerated framework that is developed from a thorough derivative analysis of camera pose in 3D Gaussian Splatting 3Dgs, CG-SLAM can perform extremely fast rendering and solve the long-standing efficiency bottleneck suffered by previous rendering-based SLAM methods.
• Figure 2: System Overview. In a 3D Gaussian field constructed from an RGB-D sequence, we can render color, depth, opacity, and uncertainty maps through a GPU-accelerated rasterizer. Additionally, we attach a new uncertainty property to each Gaussian primitive to filter informative primitives. In the mapping process, we utilize multiple rendering results to design effective loss functions towards a consistent and stable Gaussian field. Subsequently, we employ apperance and geometry cues to perform accurate and efficient tracking.
• Figure 3: Uncertainty of the Gaussian Primitives. Uncertainty of a Gaussian primitive is derived from its dominated pixels and corresponding depth biases, reflecting the geometric value and confidence of this primitive.
• Figure 4: Reconstruction Performance on Replicastraub2019replica Dataset. We qualitatively compared the mesh reconstruction results from CG-SLAM and other baselines, where CG-SLAM can produce more detailed geometry at a lower computation cost.
• Figure 5: Uncertainty Model Ablation and Anisotropy Interference. (a) Uncertainty Model Ablation. This plot illustrates that the uncertainty model significantly helps improve tracking accuracy while avoiding some extreme errors. (b) Anisotropy Interference. It can be clearly seen that in the case of w/o $\mathcal{L}_{iso}$, serious arrow-shaped artifacts occur on the edges of the image.