High-Fidelity SLAM Using Gaussian Splatting with Rendering-Guided Densification and Regularized Optimization

Shuo Sun,Malcolm Mielle,Achim J. Lilienthal,Martin Magnusson

2024-10-02

Abstract:We propose a dense RGBD SLAM system based on 3D Gaussian Splatting that provides metrically accurate pose tracking and visually realistic reconstruction. To this end, we first propose a Gaussian densification strategy based on the rendering loss to map unobserved areas and refine reobserved areas. Second, we introduce extra regularization parameters to alleviate the forgetting problem in the continuous mapping problem, where parameters tend to overfit the latest frame and result in decreasing rendering quality for previous frames. Both mapping and tracking are performed with Gaussian parameters by minimizing re-rendering loss in a differentiable way. Compared to recent neural and concurrently developed gaussian splatting RGBD SLAM baselines, our method achieves state-of-the-art results on the synthetic dataset Replica and competitive results on the real-world dataset TUM.

Robotics,Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to improve the performance of dense visual SLAM systems based on RGBD input, especially in terms of the quality of scene reconstruction and the accuracy of camera pose tracking. Specifically, the paper proposes a dense RGBD SLAM system based on 3D Gaussian Splatting, aiming to provide accurate metric pose tracking and visually realistic reconstruction. To achieve this goal, the paper mainly addresses the following key issues:

1. **Mapping of Unobserved Areas and Refinement of Observed Areas**: The paper proposes a Gaussian Splatting densification strategy based on rendering loss for mapping unobserved areas and improving the rendering quality of re - observed areas. This method can effectively fill holes in the environment and enhance the detail representation of known areas.

2. **The "Forgetting" Problem in the Continuous Mapping Process**: In the continuous mapping process, Gaussian parameters are prone to over - fitting the latest frame, resulting in a decline in the reconstruction quality of previous frames. For this reason, the paper introduces additional regularization parameters to supervise the learning process, thereby maintaining the rendering quality of previously visited areas.

3. **The Possibility of Real - Time Applications**: By online optimizing Gaussian parameters and estimating camera poses, the method in the paper can handle continuous RGBD frame inputs, eliminating the need for prior camera poses and providing the possibility for online application scenarios such as exploration.

The paper verifies the effectiveness of its method on the synthetic dataset Replica and the real - world dataset TUM - RGBD through experiments, demonstrating superior performance in terms of reconstruction quality and tracking accuracy. In addition, the paper also conducts an ablation study to prove the effectiveness of the proposed densification strategy and regularization terms.