Installation

Our simulation environment depends on and is built on top of EmbodiChain, so we need to install EmbodiChain first, and then install the RobosynChallenge extension.

Choose your setup

Based on the EmbodiChain installation guide, we offer two installation methods:

Path	Best for	Notes
Docker	First run, reproducible GPU sim	Pre-built image with CUDA 12.8, Vulkan, and Python 3.11
Local + uv	Day-to-day development	Fast installs; recommended with a virtual environment

Docker (Recommended)

Prerequisites

• Docker with NVIDIA Container Toolkit

• NVIDIA driver ≥ 535 on the host

• For GUI runs: working X11 forwarding (DISPLAY, ~/.Xauthority, /tmp/.X11-unix)

• For headless servers: no display required; use --headless in scripts

1. Construct docker container

First, please install EmbodiChain, which is the underlying simulation environment we use.

docker pull dexforce/embodichain:ubuntu22.04-cuda12.8
# The pre-configured image includes CUDA 12.8, Vulkan-related mounts, and dependencies needed for GPU simulation and rendering.

Then, open the container using the image and mount the local workspace.

mkdir RoboSynChallenge_ws && cd RoboSynChallenge_ws
git clone https://github.com/DexForce/EmbodiChain.git
cd EmbodiChain
./docker/docker_run.sh <container_name> <data_path>
# This will mount <data_path> to the /root/workspace directory in <container_name> container.
# We recommend setting <data_path> to the RoboSynChallenge_ws/ directory and placing EmbodiChain and RoboSynChallenge there.

2. Install EmbodiChain

After enter the docker container, install the EmbodiChain package:

conda create -n robosyn python=3.11
conda activate robosyn
cd /root/workspace/EmbodiChain
pip install -e . --extra-index-url http://pyp.open3dv.site:2345/simple/ --trusted-host pyp.open3dv.site
pip install "numpy<2.0"

3. Install RoboSynChallenge

Then install the RoboSynChallenge package:

cd /root/worspace
git clone https://github.com/EDEM-AI/RoboSynChallenge.git
cd RoboSynChallenge
pip install -e .

Local installation

Use a dedicated virtual environment to avoid conflicts with system Python packages.

1. Create a virtual environment

With uv (recommended):

curl -LsSf https://astral.sh/uv/install.sh | sh
uv venv --python 3.11 .venv
source .venv/bin/activate

2. Install EmbodiChain

editable install from source with uv:

mkdir RoboSynChallenge_ws && cd RoboSynChallenge_ws
git clone https://github.com/DexForce/EmbodiChain.git
cd EmbodiChain
uv venv --python 3.11 .venv && source .venv/bin/activate
uv pip install -e . \
  --extra-index-url http://pyp.open3dv.site:2345/simple/ \
  --trusted-host pyp.open3dv.site

This pulls in dexsim_engine (Python package dexsim) and the rest of the core dependencies declared in pyproject.toml.

3. Install RoboSynChallenge

Then install the RoboSynChallenge package:

cd /path/to/RoboSynChallenge_ws
git clone https://github.com/EDEM-AI/RoboSynChallenge.git
cd RoboSynChallenge
uv pip install -e .

Verify installation

1. Verify-Embodichain

python -c "import embodichain, dexsim; print('embodichain', embodichain.__version__); print('dexsim', dexsim.__version__)"

You should see version strings for both packages with no import errors.

The tutorial script scripts/tutorials/sim/create_scene.py ships with the repository. Run it from the repository root:

cd /path/to/EmbodiChain
source .venv/bin/activate
python scripts/tutorials/sim/create_scene.py

• With a display: omit --headless to open the DexSim viewer after the scene is built.

• Headless / SSH: use --headless to run without a window (FPS logs in the terminal):

python scripts/tutorials/sim/create_scene.py --headless

Press Ctrl+C to stop; the script cleans up the simulation on exit.

2 Verify-RoboSynChallenge

python -c "import robosynchallenge; print('robosynchallenge', robosynchallenge.__version__)"

You should see version strings for this package with no import errors.

You can also execute the following script to verify that the data acquisition function is working properly:

cd /path/to/RoboSynChallenge_ws/RoboSynChallenge
bash launch/run_task.sh click_bell random 2_1 --max_episodes 1

If the installation is successful, the script will start the simulation environment and display the UI of the click_bell environment. Simultaneously, it will collect one episode of data in lerobot format (version 2.1) and save it to /path/to/RoboSynChallenge/lerobot_dataset/click_bell/.

The data storage directory can be modified under the corresponding task and the corresponding gym_config settings. For example, the path to the gym_config file for the click_bell task and adding domain randomization is: /path/to/RoboSynChallenge/configs/click_bell/random/gym_config.json