UP-SLAM: Adaptively Structured Gaussian SLAM with Uncertainty Prediction in Dynamic Environments
Wancai Zheng, Linlin Ou, He Jiajie, Libo Zhou, Yan Wei, Xinyi Yu
AI summary
Problem
Existing 3D Gaussian Splatting SLAM systems suffer from poor real-time performance and reduced robustness in dynamic environments due to sequential optimization pipelines and sensitivity to moving objects.
Approach
UP-SLAM decouples tracking and mapping into a parallel framework, using a training-free uncertainty estimator to filter dynamic regions and probabilistic anchors to automatically manage Gaussian primitives.
Key results
- 59.8% improvement in localization accuracy
- 4.72 dB PSNR gain in rendering quality
- Real-time performance with reduced model size
- Open-set dynamic object filtering without semantic priors
Why it matters
Provides a robust, real-time mapping solution for robotics and AR/VR systems operating in unpredictable, dynamic real-world environments.
Abstract
Recent 3D Gaussian Splatting (3DGS) techniques for visual Simultaneous Localization and Mapping (SLAM) have significantly progressed in tracking and high-fidelity mapping. However, their sequential optimization framework and sensitivity to dynamic objects limit real-time performance and robustness in real-world scenarios. We present UP-SLAM, a real-time RGB-D SLAM system for dynamic environments that decouples tracking and mapping through a parallelized framework. A probabilistic anchor is employed to manage Gaussian primitives adaptively, enabling efficient initialization and pruning without hand-crafted thresholds. To robustly filter dynamic regions during tracking, we propose a training- free uncertainty estimator that fuses multi-modal residuals to estimate per-pixel motion uncertainty, achieving open-set dynamic object handling without reliance on semantic labels. Furthermore, a temporal encoder is designed to enhance ren- dering quality, while a shallow multilayer perception transforms low-dimensional features into DINO features, enriching the Gaussian field and enhancing uncertainty prediction robust- ness. Extensive experiments on multiple challenging datasets suggest that UP-SLAM outperforms state-of-the-art methods in both localization accuracy (by 59.8%) and rendering quality (by 4.72 dB PSNR), while maintaining real-time performance and producing reusable, artifact-free static maps in dynamic environments. The Project Page.