Note
This setup guide describes the configuration used for the original demo presented at ROSCon 2025. The demo was run on an AMD Ryzen™ AI Max+ 395 with 128GB of total RAM, 96GB of which was allocated to the GPU.
This repository is compatible with the following system:
- System: Ubuntu 24.04
- ROS 2: Jazzy, provided by pixi via RoboStack (no separate ROS 2 install needed)
- Python: 3.12
- OpenAI-compatible Endpoint for GenAI inference (llama.cpp)
You can run the demo on either a single computer or across three separate machines. Choose the guide that matches your setup.
Run HMI, agents and simulation on one computer
Important
This setup targets systems with approximately 48GB of available VRAM.
If your system does not meet this requirement, you can configure the agents to use cloud-based models instead via cloud_config.toml — see Using cloud models.
Setup from ROSCon2025, resembling a real-world scenario:
- Simulation
- HiL (Hardware-in-the-Loop)
- HMI (Human–Machine Interface)