DiskChunGS: Large-Scale 3D Gaussian SLAM Through Chunk-Based Memory Management

2025-11-28

AI Summary

Current 3D Gaussian Splatting (3DGS)-based SLAM systems are severely constrained by GPU memory, hindering scalability to large-scale scenes. To address this, we propose the first large-scale, block-based out-of-core SLAM architecture tailored for 3DGS: the scene is partitioned into spatial blocks, and a GPU-memory–disk co-scheduling mechanism is designed to retain only active blocks in GPU memory while offloading inactive ones to disk. Furthermore, SLAM pose estimation and loop closure detection are tightly integrated to ensure global consistency. Our method achieves, for the first time, end-to-end, crash-free reconstruction across all 11 KITTI sequences. Extensive evaluation on Replica, TUM-RGBD, and KITTI demonstrates superior rendering quality over state-of-the-art methods, while enabling real-time operation even on resource-constrained edge platforms such as NVIDIA Jetson.

Technology Category

• Computer Vision: 3D Computer Vision
• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Intelligent Robots: State Estimation

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Location- and context-aware Web and WoT applications and services
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines

Abstract

Recent advances in 3D Gaussian Splatting (3DGS) have demonstrated impressive results for novel view synthesis with real-time rendering capabilities. However, integrating 3DGS with SLAM systems faces a fundamental scalability limitation: methods are constrained by GPU memory capacity, restricting reconstruction to small-scale environments. We present DiskChunGS, a scalable 3DGS SLAM system that overcomes this bottleneck through an out-of-core approach that partitions scenes into spatial chunks and maintains only active regions in GPU memory while storing inactive areas on disk. Our architecture integrates seamlessly with existing SLAM frameworks for pose estimation and loop closure, enabling globally consistent reconstruction at scale. We validate DiskChunGS on indoor scenes (Replica, TUM-RGBD), urban driving scenarios (KITTI), and resource-constrained Nvidia Jetson platforms. Our method uniquely completes all 11 KITTI sequences without memory failures while achieving superior visual quality, demonstrating that algorithmic innovation can overcome the memory constraints that have limited previous 3DGS SLAM methods.

Problem

• Overcomes GPU memory limits in large-scale 3D Gaussian SLAM
• Enables globally consistent reconstruction of extensive environments
• Validates scalability on indoor, urban, and resource-limited platforms

Innovation

• Out-of-core chunk-based memory management for scalability
• Seamless integration with existing SLAM frameworks
• Validation on diverse datasets and resource-constrained platforms