Modernizing ROS 2 Skills:
Hacking and Orchestrating Cloud Brains,
Physical Sensors, and the Network


  UCL CEGE and ARC Teams

FOSDEM 2026; Track: Robotics and Simulation; Brussels / Saturday 31 January

github-repository:UCL-CyberPhysicalSystems/fosdem2026

Q1-2026 (web-animations 2025 by mxochicale)

Overview

Introduction

Modernizing ROS 2 Skills

🔧 Hacking and Orchestrating Cloud Brains, Physical Sensors, and the Network

Figure 1: Cyber physical systems, cloud layer and participants

UCL Infrastructure

Orchestrating Cloud Brains, Physical Sensors, and the Network

🤖 Orchestrating UCL infrastructure

Figure 2: Servers in UCL infrastructure and their network communication

Demos

Orchestrating Cloud Brains, Physical Sensors, and the Network

Demos: VM↔︎VM and Bare-metal↔︎VM Communication with zenoh-plugin-ros2dds

  • Docker container with ROS2 dependencies
  • Github Container Registry
  • VMs managed with terraform and kubernetes (k8s)
Figure 3: Hacking and connecting workflow

Demos: IsaacSim and IsaacLab

Figure 4: IsaacSim IDE with robot simple key control
Figure 5: Franka Lift Cube (RL Training)

Future work, Key Takeaways, Calls to Action and Acknowledgements

Future work: Multimodal Sensing

Figure 6: Fig 11 from Liu et al. 2025 Neural Brain: A Neuroscience-inspired Framework for Embodied Agents arXiv preprint: 2505.07634

Future work: Developing Embodied Agents

Figure 7: Fig 15 from Liu et al. 2025 Neural Brain: A Neuroscience-inspired Framework for Embodied Agents arXiv preprint: 2505.07634

Future events: Hamlyn Workshop 2025

  • 50 partcipants, 10 speakers, 6 posters presenters and sold out event!
  • Open-Source development in Medical Robotics Late June 2026 at London, UK
Figure 8: Participants of our 2025 event, details here

Key Takeaways & Why This Matters

  • Lowering the Barrier to Advanced Robotics at Scale

    • Cloud-ready, cost-aware infrastructure enables hands-on robotics training and research using real sensors, networks, and constraints.

  • A Proven, Scalable Model for Collaboration

    • High-performance networking and shared platforms make cross-department and cross-institution collaboration practical and repeatable.

  • A Testbed for Research, Training, and Innovation

    • The testbed supports skills transfer, research experimentation, and the rapid development of new robotics workflows.

  • Call to Action

    • Partner with us to extend the testbed, through collaboration and funding to scale training, expand research use cases, and deploy models beyond a single institution.

Sciortino et al. 2017 in Computers in Biology and Medicine https://doi.org/10.1016/j.compbiomed.2017.01.008;
He et al. 2021 in Front. Med. https://doi.org/10.3389/fmed.2021.729978

Thank You to UCL colleagues. Let’s Connect!

Extra slides

Miguel Xochicale