Fabrics

Fabrics are nonlinear, autonomous, second order differential equations that are provably stable and exhibit path consistency properties. This library brings fabrics to the GPU enabling large-scale parallelization, differentiability, and more. The intended domain is robot control and we include a manually designed fabric policy for the Kuka-Allegro robot that enabled DextrAH training.

Installation

Note: This project will download and install additional third-party open source software projects. Review the license terms of these open source projects before use.

1. Install Isaac Sim following the local conda install route.

2. Install Fabrics within your new conda env

        curl -sSL https://install.python-poetry.org | python3 - --version 1.8.3
        git lfs clone [email protected]:NVlabs/FABRICS.git
        cd <FABRICS>
        poetry install
        or
        python -m pip install -e .

3. Patch urdfpy dependency networkx

       cd <fabrics_dir>
       chmod +x urdfpy_patch.sh
       ./urdfpy_patch.sh

4. Ensure high enough GLIBCXX_ can be found

        conda install -c conda-forge libstdcxx-ng
        conda install -c conda-forge libgcc-ng=12 libstdcxx-ng=12

Examples

Once installed, you should be able to run the example script in the example directory, e.g.:

Kuka-Allegro (DextrAH) fabric:

python <fabrics_dir>/examples/kuka_allegro_pose_fabric_example.py --batch_size=10 --render --cuda_graph

Notes

One can update dependences in deps.txt file, remove pyproject.toml and poetry.lock files, and regenerate them

    cd <fabrics_dir>
    rm pyproject.toml poetry.lock
    poetry init --name "fabrics_sim" --no-interaction
    xargs poetry add < deps.txt

Add the following after authors in pyproject.toml

    packages = [ 
    { include = "fabrics_sim", from = "src" }
    ]

Install the project

    poetry install