Splat-LOAM: Gaussian Splatting LiDAR Odometry and Mapping
This work addresses the challenge of efficient scene representation for real-time robotics estimation tasks, though it is incremental as it builds on existing Gaussian Splatting methods.
The paper tackled the problem of accurate and lightweight LiDAR-based ego-motion estimation and mapping by developing a pipeline using Gaussian Splatting primitives, achieving state-of-the-art mapping results with minimal GPU requirements.
LiDARs provide accurate geometric measurements, making them valuable for ego-motion estimation and reconstruction tasks. Although its success, managing an accurate and lightweight representation of the environment still poses challenges. Both classic and NeRF-based solutions have to trade off accuracy over memory and processing times. In this work, we build on recent advancements in Gaussian Splatting methods to develop a novel LiDAR odometry and mapping pipeline that exclusively relies on Gaussian primitives for its scene representation. Leveraging spherical projection, we drive the refinement of the primitives uniquely from LiDAR measurements. Experiments show that our approach matches the current registration performance, while achieving SOTA results for mapping tasks with minimal GPU requirements. This efficiency makes it a strong candidate for further exploration and potential adoption in real-time robotics estimation tasks.