Installation & Deployment Issues

[Nan-Tetheria]

Hi,

Thanks for the great work! Unfortunately, the code is currently not installable or deployable on my end.

GPU: NVIDIA GeForce RTX 4070 Laptop GPU

CUDA: CUDA: 12.4 (V12.4.131)

Python: 3.10.12

OS: Ubuntu 22.04.4 LTS

🧩 Local Installation Attempt

I first tried installing locally following the instructions.

The dependency list seems incomplete — several required packages were missing, including the wheel build dependencies.

After manually fixing those (with AI assistance), running:

pip install -e .[full] --no-build-isolation

caused 100% CPU usage across all cores and hung indefinitely.
This behavior was consistent on multiple machines and servers, suggesting a possible issue with the build configuration or pyproject.toml.

🐳 Docker Build Attempt

Next, I tried to reproduce the setup using Docker, but ran into similar problems.

It’s unclear where to place the IsaacLab directory, leading to repeated path and directory errors.

Even after fixing that, PyTorch version conflicts arose between IsaacLab / IsaacSim and Torch3D, making it hard to produce a working image.

Overall, the process is quite fragile and difficult to reproduce.

💡 Suggestions

To make the project more accessible and reproducible, it would be great if you could:

Provide a prebuilt Docker image (e.g., on Docker Hub or as a downloadable .tar), or

Update the documentation with clearer steps for environment setup, particularly:

Exact dependency versions

Expected location of IsaacLab

Recommended PyTorch/CUDA combinations

I would be happy to test or verify once a fix or image is available.
Thank you again for sharing this work — looking forward to getting it running! 🚀

Nan

[renezurbruegg]

Thanks for the feedback - I am sorry to hear that you encounter issues with the setup.

Can you share the log of the docker image build process? I set up the docker image specifically to ensure that there would not be a conflict with pytorch3d. It is also the image I use on a regular basis.