Install & Download¶

1. Dependencies¶

System Support:

We currently best support Linux based systems. There is limited support for windows and no support for MacOS at the moment. We are working on trying to support more features on other systems but this may take some time. Most constraints stem from what the SAPIEN package is capable of supporting.

System / GPU	CPU Sim	GPU Sim	Rendering
Linux / NVIDIA GPU	✅	✅	✅
Windows / NVIDIA GPU	✅	❌	✅
Windows / AMD GPU	✅	❌	✅
WSL / Anything	✅	❌	❌
MacOS / Anything	✅	❌	✅

Occasionally, data collection may get stuck when using A/H series GPUs. This issue may be related to RoboTwin issue #83 and SAPIEN issue #219.

Python versions:

Python 3.10

CUDA version:

12.1 (Recommended)

Hardware:

Rendering: NVIDIA or AMD GPU
Ray tracing: NVIDIA RTX GPU or AMD equivalent
Ray-tracing Denoising: NVIDIA GPU
GPU Simulation: NVIDIA GPU

Software:

Ray tracing: NVIDIA Driver >= 470
Denoising (OIDN): NVIDIA Driver >= 520

1.1 Additional Requirements for Docker¶

When running in a Docker container, ensure that the following environment variable is set when starting the container:

-e NVIDIA_DRIVER_CAPABILITIES=compute,utility,graphics

Important : The graphics capability is essential. Omitting it may result in segmentation faults due to missing Vulkan support.

For more information, see HERE.

2. Install Vulkan (if not installed)¶

sudo apt install libvulkan1 mesa-vulkan-drivers vulkan-tools

Check by running vulkaninfo

3. Basic Env¶

First, prepare a conda environment.

conda create -n RoboTwin python=3.10 -y
conda activate RoboTwin

RoboTwin 2.0 Code Repo: https://github.com/RoboTwin-Platform/RoboTwin

git clone https://github.com/RoboTwin-Platform/RoboTwin.git

Then, run script/_install.sh to install basic envs and CuRobo:

bash script/_install.sh

If you meet curobo config path issue, try to run python script/update_embodiment_config_path.py

If you encounter any problems, please refer to the manual installation section. If you are not using 3D data, a failed installation of pytorch3d will not affect the functionality of the project.

If you haven't installed ffmpeg, please turn to https://ffmpeg.org/. Check it by running ffmpeg -version.

4. Download Assets (RoboTwin-OD, Texture Library and Embodiments)¶

To download the assets, run the following command. If you encounter any rate-limit issues, please log in to your Hugging Face account by running huggingface-cli login:

bash script/_download_assets.sh

The structure of the assets folder should be like this:

assets
├── background_texture
├── embodiments
│   ├── embodiment_1
│   │   ├── config.yml
│   │   └── ...
│   └── ...
├── objects
└── ...

5. Manual Installation (Only when step 3 failed)¶

Install requirements
```
pip install -r requirements.txt
```

Install pytorch3d

pip install "git+https://github.com/facebookresearch/pytorch3d.git@stable"

Install CuRobo

cd envs
git clone https://github.com/NVlabs/curobo.git
cd curobo
pip install -e . --no-build-isolation
cd ../..

Adjust code in mplib (Important)
You can use pip show mplib to find where the mplib installed.
Remove or collide

# mplib.planner (mplib/planner.py) line 807
# remove `or collide`

if np.linalg.norm(delta_twist) < 1e-4 or collide or not within_joint_limit:
                return {"status": "screw plan failed"}
=>
if np.linalg.norm(delta_twist) < 1e-4 or not within_joint_limit:
                return {"status": "screw plan failed"}