GLAM-SLAM: Real-time Gaussian Large-scale Mapping via Flow Densification and Spatial Decomposition

2026-07-23

AI Summary

This work addresses the limitations of existing monocular Gaussian-point-based SLAM systems in large-scale outdoor environments, which suffer from poor real-time performance, high GPU memory consumption, and restricted sequence length. To overcome these challenges, we propose GLAM-SLAM, a decoupled architecture comprising a lightweight feature-based front-end for robust camera tracking and a novel back-end that leverages a sparse anchor grid with spatial partitioning. The back-end efficiently initializes 3D Gaussians using an epipolar geometry-constrained optical flow densification strategy and enhances reconstruction consistency by injecting spatial inductive bias through a multilayer perceptron (MLP). Evaluated on long-sequence datasets including KITTI, Oxford RobotCar, and Málaga, GLAM-SLAM achieves real-time performance while improving reconstruction quality by 15% over state-of-the-art methods, demonstrating strong scalability to large-scale scenes.

Abstract

Existing Gaussian-splatting-based monocular Simultaneous Localization and Mapping (SLAM) systems are either tailored to short sequences, are not real-time, or suffer from prohibitive GPU memory requirements, limiting their applicability in realistic, long-horizon scenarios. To address this, we present GLAM-SLAM, a real-time, decoupled Gaussian-splatting SLAM system designed for large-scale outdoor scenes. We ensure lightweight tracking using a robust, feature-based SLAM frontend, while for mapping, we adopt a structured, sparse anchor grid representation that ensures scalable operation and maintains scene coherence across long-term sequences. To satisfy the dense initialization requirements of 3D Gaussian Splatting (3DGS), we introduce a geometry-based flow-densification anchoring strategy using epipolar constraints. Furthermore, by treating mapping as a multi-scene problem, we propose a scene-partitioning strategy that introduces a strong spatial inductive bias via MLP initializations to generate localized Gaussians. We evaluate our system on the challenging, long-sequence KITTI Odometry, Oxford RobotCar, and M'alaga datasets. Extensive ablations and comparisons demonstrate a 15% improvement in reconstruction quality over the second-best performer, while maintaining real-time performance and the ability to scale to longer sequences. Code is publicly available for the benefit of the community.

Problem

• Gaussian Splatting
• SLAM
• large-scale mapping
• real-time
• GPU memory

Innovation

• Gaussian Splatting
• Real-time SLAM
• Flow Densification
• Spatial Decomposition
• Large-scale Mapping