Search papers, labs, and topics across Lattice.
This paper introduces VarSplat, a novel 3D Gaussian Splatting (3DGS)-SLAM system that explicitly models per-splat appearance variance to improve robustness in challenging scenarios. By propagating splat-level uncertainty to a differentiable per-pixel uncertainty map via alpha compositing, VarSplat guides tracking, submap registration, and loop detection to focus on reliable regions. Experiments on synthetic and real-world datasets demonstrate that VarSplat achieves competitive or superior performance in tracking, mapping, and novel view synthesis compared to existing dense RGB-D SLAM methods.
By explicitly modeling per-splat appearance variance, VarSplat enables more robust 3D Gaussian Splatting SLAM, particularly in low-texture or reflective environments where existing methods struggle.
Simultaneous Localization and Mapping (SLAM) with 3D Gaussian Splatting (3DGS) enables fast, differentiable rendering and high-fidelity reconstruction across diverse real-world scenes. However, existing 3DGS-SLAM approaches handle measurement reliability implicitly, making pose estimation and global alignment susceptible to drift in low-texture regions, transparent surfaces, or areas with complex reflectance properties. To this end, we introduce VarSplat, an uncertainty-aware 3DGS-SLAM system that explicitly learns per-splat appearance variance. By using the law of total variance with alpha compositing, we then render differentiable per-pixel uncertainty map via efficient, single-pass rasterization. This map guides tracking, submap registration, and loop detection toward focusing on reliable regions and contributes to more stable optimization. Experimental results on Replica (synthetic) and TUM-RGBD, ScanNet, and ScanNet++ (real-world) show that VarSplat improves robustness and achieves competitive or superior tracking, mapping, and novel view synthesis rendering compared to existing studies for dense RGB-D SLAM.
