Setup

Requirements

Prepare the host machine before starting Cyclo Intelligence.

Requirement	How to check
Docker 24 or later with Docker Compose	docker compose version
Free disk space for Docker images, datasets, and models	Keep at least 35 GB available before the first start.
Available UI port	Make sure port 7080 is free, or set CYCLO_UI_PORT to another port before starting.
NVIDIA Container Toolkit	Required for GPU-based policy containers. The main UI and data tools can run without a GPU when the no-GPU compose override is used.

Cyclo Intelligence uses public Docker images by default, so Docker Hub login is not required.

Install

Install Cyclo Intelligence on the machine where the web UI and Docker containers will run.

curl -fsSL https://raw.githubusercontent.com/ROBOTIS-GIT/cyclo_intelligence/main/install.sh | bash

For an existing checkout, update the repository and submodules manually:

cd $HOME/cyclo_intelligence
git pull
git submodule update --init --recursive

Start The Container

Start the main Cyclo Intelligence container from the repository root:

cd $HOME/cyclo_intelligence
./docker/container.sh start

Then enter the container:

./docker/container.sh enter

Start Zenoh

Cyclo Intelligence uses Zenoh-based ROS 2 communication for all supported robots.

important

Before launching the Cyclo Intelligence ROS 2 process, start the Zenoh daemon for the robot network.

Follow the AI Worker Zenoh Communication guide for the robot you are setting up.

Set ZENOH_CONFIG_OVERRIDE in ~/.bashrc inside the Cyclo Intelligence container according to where the Zenoh daemon is running. If ZENOH_CONFIG_OVERRIDE already exists in ~/.bashrc, update the existing line instead of adding another one.

When Cyclo Intelligence and the Zenoh daemon run on the same machine:

echo "export ZENOH_CONFIG_OVERRIDE='transport/shared_memory/enabled=true'" >> ~/.bashrc
source ~/.bashrc

To use Cyclo Intelligence on an external PC that is not running the Zenoh daemon, set Cyclo Intelligence to client mode and use the IP address of the PC where the Zenoh daemon is running as the endpoint:

echo "export ZENOH_CONFIG_OVERRIDE='transport/shared_memory/enabled=true;mode=\"client\";connect/endpoints=[\"tcp/192.168.0.42:7447\"]'" >> ~/.bashrc
source ~/.bashrc

Replace 192.168.0.42 with the actual Zenoh daemon IP address.

Launch Cyclo Intelligence

Inside the container, run either command:

cyclo_intelligence

or:

ros2 launch orchestrator cyclo_intelligence_bringup.launch.py

The cyclo_intelligence alias runs the same ROS 2 launch command.

After the launch is running, open the web UI:

http://localhost:7080/

If Cyclo Intelligence is running on another machine, replace localhost with that machine's IP address.

For example, if Cyclo Intelligence is running on a machine with IP address 192.168.0.42, open:

http://192.168.0.42:7080/

For AI Worker, you can also use the robot hostname when it is available on the network. For example:

http://ffw-{serial-number}.local:7080/