BayesianGS-SLAM improves camera tracking with uncertainty aware rendering
BayesianGS-SLAM: Uncertainty-Aware Neural Rendering SLAM via Probabilistic Formulation
Robotics
Summary
Mapping environments with cameras can be tricky when the system isn't sure how accurate its visual clues are. The authors developed BayesianGS-SLAM, a method that estimates how uncertain its color and depth predictions are during 3D scene mapping. This helps the system avoid mistakes when deciding how the camera moves and which parts of the scene need updating. Their approach uses these uncertainty estimates throughout tracking and mapping, leading to better performance and fewer unnecessary updates.
What this means in practice
- •For robotics engineers: Enhance robot navigation systems by integrating BayesianGS-SLAM for more reliable camera pose tracking under uncertain visual inputs.
- •For augmented reality developers: Improve AR device mapping efficiency and visual stability by using uncertainty-aware keyframe selection with BayesianGS-SLAM.
Authors
Kyeongsu Kang, Seongbo Ha, Sibaek Lee, Hyeonwoo Yu
Abstract
Neural-rendering-based SLAM relies on rendered RGB-D residuals for camera tracking and map optimization, but the reliability of these predictions can vary substantially because of sensor noise, limited observation coverage, and incomplete map representations. Without an explicit reliability estimate, unreliable residuals may adversely affect pose optimization, while frames already well explained by the current map may trigger redundant mapping updates. In this paper, we present BayesianGS-SLAM, an uncertainty-aware 3D Gaussian Splatting SLAM framework that estimates predictive color and depth uncertainty during mapping and consistently reuses it across the SLAM pipeline. Our tractable probabilistic formulation combines a sensor-noise uncertainty component with an opacity-induced map-representation component propagated through the rendering process. The resulting predictive uncertainty is used to augment mapping, normalize tracking residuals through a robust pose objective, and evaluate incoming frames using a predictive-surprise-based keyframe criterion. Unlike prior uncertainty-aware neural-rendering SLAM methods that primarily consider color uncertainty or use uncertainty only during mapping, our framework estimates predictive uncertainty for both color and depth and integrates it into mapping, tracking, and keyframe selection. Evaluations on real-world RGB-D datasets demonstrate substantially improved depth uncertainty-error ranking compared with existing uncertainty-aware SLAM methods. Moreover, the proposed keyframe-selection strategy reduces the number of selected keyframes and mapping calls while maintaining competitive tracking and rendering performance.