Spatial geometry improves 3D camera tracking for outdoor navigation
SG^2SLAM: Spatial Geometry-Informed 3D Gaussian Splatting SLAM for Monocular Outdoor Navigation
Robotics
Summary
Estimating the position of a moving camera and creating a detailed 3D map using only one camera is challenging, especially outdoors. The authors present a new method called SG^2SLAM that improves this by using a spatial geometry approach, which looks at larger chunks of data rather than individual frames to reduce errors. They also use a technique that accounts for different camera angles to better integrate the 3D shapes into the map. Tested on real-world driving datasets, their method provides more accurate tracking and mapping than previous methods.
What this means in practice
- •For autonomous vehicle developers: Improve real-time 3D mapping and camera pose estimation for self-driving cars navigating outdoor environments with monocular cameras.
- •For robotics navigation teams: Enable more accurate and globally consistent navigation maps from single-camera robots operating outdoors by reducing tracking errors in SLAM.
Authors
Ziang Guo, Didar Seyidov, Yixiao Zhou, Xuefeng Zhang, Ali Alridha Abdulkarim, Chen Min, Zufeng Zhang, Dzmitry Tsetserukou
Abstract
Monocular outdoor navigation requires robust camera motion estimation and a dense, navigable map, where 3D Gaussian Splatting SLAM serves this task through an explicit differentiable parameterization and closed-form projection. However, typically, the geometry is recovered by associating 2D pixels with 3D points, so a modular SLAM system absorbs tracking error into the map, leading to cascading degradation. To address these gaps, we present SG^2SLAM, which couples streaming spatial geometry with a ray-conditioned 3DGS kernel. Streaming spatial geometry replaces the per- frame cross-dimensional correspondence with a chunk-level metric prior that supervises tracking and mapping from one source. A ray-conditioned kernel extends each primitive with a viewing-direction subspace that senses the monocular camera viewpoints. The joint optimization over spatial geometry and poses ensures a globally consistent map and trajectory outcome. Experiments on the KITTI and Waymo Open datasets show that our SG^2SLAM surpasses the per-frame association baselines in the monocular outdoor navigation task. Code will be made publicly available upon acceptance.