DUDG-SLAM: Dynamic Replay and Depth-Uncertainty-Guided Gaussian SLAM for Robust RGB-D Reconstruction

Abstract

Highlights What are the main findings? A dynamic historical keyframe replay strategy is proposed to reintroduce high-value historical views based on reconstruction error, forgetting degree, and visibility variation. An RGB-D-prioritized uncertainty-weighted depth supervision mechanism is proposed to complement unreliable sensor depth with filtered monocular depth. What are the implications of the main findings? The proposed replay strategy maintains long-term supervision for previously mapped regions, thereby alleviating texture drift and local map degradation in RGB-D 3DGS-SLAM. The proposed depth supervision mechanism provides more stable geometric constraints in regions with missing or unreliable depth, reducing floating Gaussians, blurred boundaries, and local geometric distortions while improving reconstruction quality under RGB-D observation conditions. Abstract RGB-D SLAM systems based on 3D Gaussian Splatting (3DGS) often suffer from map degradation caused by diminishing historical supervision, noisy depth observations, and local-window optimization during online reconstruction. To address these issues, we propose DUDG-SLAM, a Dynamic Replay and Depth-Uncertainty-Guided Gaussian SLAM framework. The proposed method selectively replays historical keyframes according to reconstruction error, forgetting degree, and Gaussian visibility, thereby restoring supervision in previously reconstructed regions. In addition, RGB-D sensor depth is retained as the primary geometric constraint, while scale-aligned Depth Pro predictions are introduced only in regions with missing or unreliable sensor depth. A pixel-wise uncertainty-weighted log-depth loss is further designed to reduce the influence of noisy depth observations and unreliable depth boundaries. Experiments on Replica and TUM RGB-D show that DUDG-SLAM improves rendering quality, trajectory accuracy, and reconstruction robustness. Compared with the baseline, DUDG-SLAM consistently improves PSNR and SSIM while reducing LPIPS and ATE.