CyberPhysics

CyberPhysics is a platform for deploying robotic and ML applications.

Features

• browser-based graphical user interface
• access to latest ROS tools
• deployable to kubernetes
• and more...

Installation

(This project is designed for Ubuntu 24.04)

./install.sh

Building images

From the applications folder run "make build_<app_name>". For example:

cd applications && make build_ros2

Usage

Write new applications and put them in the applications folder.

Compose your applications together using docker compose in the compositions folder.

Launch it with docker compose. For example:

docker compose -f compositions/jupyter.yaml up

For viewing ROS data, run the Foxglove compose file, and launch Foxglove on your host machine.

foxglove-studio

A web based GUI can be built with NiceGUI, and then accessed through your host machine's browser:
http://localhost:8080/

Nvblox extension for Foxglove (on host)

Reference:
https://github.com/NVIDIA-ISAAC-ROS/isaac_ros_nvblox/tree/main/nvblox_foxglove

1. git clone https://github.com/NVIDIA-ISAAC-ROS/isaac_ros_nvblox.git
2. cd nvblox_foxglove
3. (install npm==20.10.0, see below)
4. npm install
5. npm run local-install

To install npm==20.10.0:

curl https://raw.githubusercontent.com/creationix/nvm/master/install.sh | bash
nvm install 20.10.0
nvm use 20.10.0

Examples

ARDUINO:

• Connect Arduino Teensy 4.1 to USB port on host computer.
• Launch "foxglove-studio" from command line.

cd compositions
docker compose -f arduino.yaml up

• Messages from the Arduino board should now be visible

SPATIAL RECONSTRUCTION:

• Connect Realsense 435i to USB port on host computer.

cd compositions  
docker compose -f reconstruction.yaml up  

• Launch "foxglove-studio" from command line.
• Select "Open connection" -> "Rosbridge" -> "Open"
• Messages from the Nvblox board should now be visible

Documentation

Have a look at the README in each application's folder for explanations of what they do.

Why?

Many cyber-physical systems quickly become unamangeable as complexity and dependency conflicts scale exponentially with the number of components. This project should help with that.

Note on real time performance

Real time performance can be achieved by pinning a process to a core:

docker run --cpuset-cpus="0,1" -it your_image your_command

services:
  your_service:
    image: your_image
    cpuset: "0,1"

Then reserve the cores needed before running the containers:

cset shield --cpu 0,1 --kthread on 

Then set cgroup-parent in dockerd config "/etc/docker/daemon.json":

{
  "cgroup-parent": "/system"
}

Then to restore the host system:

cset shield --reset