Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

ROSCell: A ROS2-Based Framework for Automated Formation and Orchestration of Multi-Robot Systems

Jiangtao Shuai, Marvin Carl May, Sonja Schimmler, Manfred Hauswirth

Abstract

Modern manufacturing under High-Mix-Low-Volume requirements increasingly relies on flexible and adaptive matrix production systems, which depend on interconnected heterogeneous devices and rapid task reconfiguration. To address these needs, we present ROSCell, a ROS2-based framework that enables the flexible formation and management of a computing continuum across various devices. ROSCell allows users to package existing robotic software as deployable skills and, with simple requests, assemble isolated cells, automatically deploy skill instances, and coordinate their communication to meet task objectives. It provides a scalable and low-overhead foundation for adaptive multi-robot computing in dynamic production environments. Experimental results show that, in the idle state, ROSCell substantially reduces CPU, memory, and network overhead compared to K3s-based solutions on edge devices, highlighting its energy efficiency and cost-effectiveness for large-scale deployment in production settings. The source code, examples, and documentation will be provided on Github.

ROSCell: A ROS2-Based Framework for Automated Formation and Orchestration of Multi-Robot Systems

Abstract

Modern manufacturing under High-Mix-Low-Volume requirements increasingly relies on flexible and adaptive matrix production systems, which depend on interconnected heterogeneous devices and rapid task reconfiguration. To address these needs, we present ROSCell, a ROS2-based framework that enables the flexible formation and management of a computing continuum across various devices. ROSCell allows users to package existing robotic software as deployable skills and, with simple requests, assemble isolated cells, automatically deploy skill instances, and coordinate their communication to meet task objectives. It provides a scalable and low-overhead foundation for adaptive multi-robot computing in dynamic production environments. Experimental results show that, in the idle state, ROSCell substantially reduces CPU, memory, and network overhead compared to K3s-based solutions on edge devices, highlighting its energy efficiency and cost-effectiveness for large-scale deployment in production settings. The source code, examples, and documentation will be provided on Github.
Paper Structure (12 sections, 4 equations, 5 figures, 1 table)

This paper contains 12 sections, 4 equations, 5 figures, 1 table.

Table of Contents

  1. INTRODUCTION
  2. RELATED WORK
  3. Approach
  4. Features and Node Roles
  5. Communication Pattern and Cell Formation
  6. Skill Model and Orchestration
  7. Skill Deployment and Lifecycle Management
  8. Evaluation
  9. System Overhead
  10. Network Throughput
  11. Multi-Object Pose Estimation
  12. CONCLUSIONS

Figures (5)

  • Figure 1: Workflow of ROSCell. Dashed lines show I/O data flows and solid lines information flows. Stacked blocks indicate possible multiple nodes. The gray block illustrates that cell nodes can be any type of resource in the device-edge-cloud continuum. The numbered steps show the process of cell formation and software deployment: (1) Developers wrap software into ROSCell; (2) The production controller enables devices as ROSCell nodes; (3)&(4) The coordination node forms its cells via the instruction from the production controller; (5) The operator sends tasks to the coordination node; (6) ROSCell orchestrates and executes the skill instances within the cell.
  • Figure 2: Class diagram of ROSCell message handling mechanism. The circle with C represents a class, while the circle with I represents an interface.
  • Figure 3: Class diagram of the skill descriptor hierarchy, with an object detection skill as an example. Two implementation models ( yolo_v6 and pointpillars ) are shown, each offering CPU and GPU deployment options.
  • Figure 4: Comparison of packet and traffic loads
  • Figure 5: Multi-object pose estimation experiments: a) illustrates an example use case in which robot poses are estimated from AprilTags by an edge device connected to the camera. The edge device functions as a ROSCell cell member, with the camera-stream ROS2 node and the pose-estimation skill instance running in two separate Docker containers. b) presents the per-frame processing time of the estimator under different deployment strategies.