WildGS-SLAM: Monocular Gaussian Splatting SLAM in Dynamic Environments
This addresses the challenge of robust SLAM for robotics or AR/VR applications in real-world dynamic settings, representing a domain-specific incremental improvement.
The paper tackles the problem of monocular SLAM in dynamic environments by developing WildGS-SLAM, which integrates uncertainty-aware geometric mapping to handle moving objects. Results show it achieves artifact-free view synthesis and outperforms state-of-the-art methods on multiple datasets.
We present WildGS-SLAM, a robust and efficient monocular RGB SLAM system designed to handle dynamic environments by leveraging uncertainty-aware geometric mapping. Unlike traditional SLAM systems, which assume static scenes, our approach integrates depth and uncertainty information to enhance tracking, mapping, and rendering performance in the presence of moving objects. We introduce an uncertainty map, predicted by a shallow multi-layer perceptron and DINOv2 features, to guide dynamic object removal during both tracking and mapping. This uncertainty map enhances dense bundle adjustment and Gaussian map optimization, improving reconstruction accuracy. Our system is evaluated on multiple datasets and demonstrates artifact-free view synthesis. Results showcase WildGS-SLAM's superior performance in dynamic environments compared to state-of-the-art methods.