Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Self-Organizing Edge Computing Distribution Framework for Visual SLAM

Jussi Kalliola, Lauri Suomela, Sergio Moreschini, David Hästbacka

TL;DR

This work presents a self-organizing distributed Visual SLAM framework that distributes SLAM modules across a network of heterogeneous devices or runs standalone on a single device. It introduces a three-layer architecture (core SLAM, distribution/state management, and ROS2/FastDDS-based communication) built around monocular ORB SLAM3, and a heuristic distribution policy to enable collaboration with resilience to network faults. Empirical results on EuRoC, TUM, and real-office data show the distributed system achieves comparable accuracy and resource utilization to a monolithic baseline, while enabling multi-node execution and graceful degradation when connectivity is lost. The study highlights state-management challenges in distributed SLAM and outlines future directions to enhance initialization and global-map consistency for robust real-world deployment.

Abstract

Localization within a known environment is a crucial capability for mobile robots. Simultaneous Localization and Mapping (SLAM) is a prominent solution to this problem. SLAM is a framework that consists of a diverse set of computational tasks ranging from real-time tracking to computation-intensive map optimization. This combination can present a challenge for resource-limited mobile robots. Previously, edge-assisted SLAM methods have demonstrated promising real-time execution capabilities by offloading heavy computations while performing real-time tracking onboard. However, the common approach of utilizing a client-server architecture for offloading is sensitive to server and network failures. In this article, we propose a novel edge-assisted SLAM framework capable of self-organizing fully distributed SLAM execution across a network of devices or functioning on a single device without connectivity. The architecture consists of three layers and is designed to be device-agnostic, resilient to network failures, and minimally invasive to the core SLAM system. We have implemented and demonstrated the framework for monocular ORB SLAM3 and evaluated it in both fully distributed and standalone SLAM configurations against the ORB SLAM3. The experiment results demonstrate that the proposed design matches the accuracy and resource utilization of the monolithic approach while enabling collaborative execution.

Self-Organizing Edge Computing Distribution Framework for Visual SLAM

TL;DR

This work presents a self-organizing distributed Visual SLAM framework that distributes SLAM modules across a network of heterogeneous devices or runs standalone on a single device. It introduces a three-layer architecture (core SLAM, distribution/state management, and ROS2/FastDDS-based communication) built around monocular ORB SLAM3, and a heuristic distribution policy to enable collaboration with resilience to network faults. Empirical results on EuRoC, TUM, and real-office data show the distributed system achieves comparable accuracy and resource utilization to a monolithic baseline, while enabling multi-node execution and graceful degradation when connectivity is lost. The study highlights state-management challenges in distributed SLAM and outlines future directions to enhance initialization and global-map consistency for robust real-world deployment.

Abstract

Localization within a known environment is a crucial capability for mobile robots. Simultaneous Localization and Mapping (SLAM) is a prominent solution to this problem. SLAM is a framework that consists of a diverse set of computational tasks ranging from real-time tracking to computation-intensive map optimization. This combination can present a challenge for resource-limited mobile robots. Previously, edge-assisted SLAM methods have demonstrated promising real-time execution capabilities by offloading heavy computations while performing real-time tracking onboard. However, the common approach of utilizing a client-server architecture for offloading is sensitive to server and network failures. In this article, we propose a novel edge-assisted SLAM framework capable of self-organizing fully distributed SLAM execution across a network of devices or functioning on a single device without connectivity. The architecture consists of three layers and is designed to be device-agnostic, resilient to network failures, and minimally invasive to the core SLAM system. We have implemented and demonstrated the framework for monocular ORB SLAM3 and evaluated it in both fully distributed and standalone SLAM configurations against the ORB SLAM3. The experiment results demonstrate that the proposed design matches the accuracy and resource utilization of the monolithic approach while enabling collaborative execution.
Paper Structure (28 sections, 1 equation, 5 figures, 1 table)

This paper contains 28 sections, 1 equation, 5 figures, 1 table.

Table of Contents

  1. INTRODUCTION
  2. RELATED WORK
  3. Visual SLAM
  4. Collaborative SLAM
  5. Edge-assisted SLAM
  6. DISTRIBUTED VSLAM FRAMEWORK
  7. Initialization
  8. Connectivity
  9. State management model
  10. Distribution Policy
  11. Requirements for SLAM distribution
  12. DISTRIBUTED MONOCULAR ORB SLAM3
  13. Core: ORB SLAM3
  14. Distribution layer
  15. Communication Layer
  16. ...and 13 more sections

Figures (5)

  • Figure 1: High-level illustration of the proposed distribution framework for VSLAM. The system in multi-node (bottom) and single-node configuration (top).
  • Figure 2: Connectivity scheme. Arrows indicate the direction of the data flow and its corresponding ROS2 topic.
  • Figure 3: Distribution system architecture.
  • Figure 4: State managers for Keyframes and Maps. A dashed outline indicates that data is transmitted to another module/layer.
  • Figure 5: Artifacts can be seen between timesteps 200-220. KeyFrames and full trajectory plotted against ground truth. EuRoC, MH_03 dataset.