Facing issue while installing Isaac-GR00T on Nvidia Agx Orin board

Robotics & Edge Computing Jetson & Embedded Systems Jetson AGX Orin
1

Hi,
I’m trying to install and run Isaac-GR00T on Nvidia Agx Orin by the steps mentioned in this link. After bash deployment_scripts/setup_env.sh command I’m encountering an error as shown in the below image.

Note: I’m trying to install using normal installation method, not Docker.

System:Nvidia Agx Orin 64GB Developer Kit
Jetpack: 6.2.1

image
image1598×850 140 KB

2

*** Please note that this reply is generated by LLM automatically ***

I’d be happy to help you with your question, but it seems like the question is incomplete and doesn’t match the provided context. However, I’ll try to provide a comprehensive answer based on the context.

The context appears to be related to NVIDIA Jetson devices, specifically the Jetson AGX Orin and Jetson Orin Nano, and various issues with containers, Docker, and installations.

Regarding the installation of IsaacGR00T on the NVIDIA AGX Orin, I couldn’t find any specific information in the provided context. However, I can offer some general guidance.

IsaacGR00T is likely a custom application or framework, and the installation process may vary depending on the specific requirements and dependencies. Since you’re trying to install it using the normal installation method, not Docker, you may need to follow the instructions provided in the link you mentioned (which is not included in the context).

Some general tips for installing applications on the Jetson AGX Orin:

  1. Ensure you have the correct dependencies installed, including any required libraries or frameworks.
  2. Follow the installation instructions carefully, and make sure you’re running the correct commands with the correct permissions.
  3. If you encounter errors, try to troubleshoot the issue by checking the logs and error messages.

Comparing the installation process on the Jetson AGX Orin to other devices, such as the Jetson Orin Nano, may not be directly applicable, as the hardware and software configurations may differ. However, some general principles, such as ensuring correct dependencies and following installation instructions, remain the same.

If you could provide more information about the error you’re encountering and the specific installation instructions you’re following, I may be able to offer more specific guidance.

*** Please give the thumbs up if you get answers you like. Or provide feedback to help us improve the answer. ***

5

Hi,

Could you share the complete log with us?
We didn’t meet the issue when running the command:

$ bash deployment_scripts/setup_env.sh
...
byte-compiling /home/nvidia/.local/lib/python3.10/site-packages/decord/data/__init__.py to __init__.cpython-310.pyc
byte-compiling /home/nvidia/.local/lib/python3.10/site-packages/decord/__init__.py to __init__.cpython-310.pyc
running install_data
creating /home/nvidia/.local/decord
copying ../build/libdecord.so -> /home/nvidia/.local/decord
running install_egg_info
Copying decord.egg-info to /home/nvidia/.local/lib/python3.10/site-packages/decord-0.6.0-py3.10.egg-info
running install_scripts

Thanks.

6

Hi @AastaLLL Thanks for your response. Please find the below attached log file.

gr00t_log1.txt (75.9 KB)

7

Hi,

Just double-confirm that we can compile ffmeg without error.
Could you try it manually?

$ git clone https://git.ffmpeg.org/ffmpeg.git
$ cd ffmpeg
$ git checkout n4.4.2
$ ./configure \
  --enable-shared \
  --enable-pic \
  --prefix=/usr
$ make -j$(nproc)
$ sudo make install

Thanks.

8

Hi @AastaLLL I was on leave, so I’m replying late. I’ve tried these steps, and they work fine. What should I do next?
Thanks.

9

Hi,

The error you shared is related to the ffmpeg compilation.
Since you can build it correctly, please try to remove he ffmpeg part in the script below and run it again.

Thanks.

10

After making changes in the setup_env.sh file, I’m also getting some errors, which I’ve attached in the text file.

terminal_log.txt (16.6 KB)

image
image1330×891 29.1 KB

11

Hi,

This looks like a known issue related to the setuptools.
Could you try the suggestion shared below to see if it can work?

Thanks.

12

Thanks for your response @AastaLLL . My error has been resolved by downgrading setuptools.
pip install setuptools==68

13

image
image1714×837 57.4 KB

Hi @AastaLLL , Installation using the bash deployment_scripts/setup_env.sh command was successful. However, when I try to run python deployment_scripts/gr00t_inference.py --inference-mode=pytorch for inference, I encounter a module import error. I noticed that PyTorch is not installed in this environment. When I try to install it using pip install torch, I get a few more module import errors. I think some dependencies have not been installed yet.

14

Hi,

Please try to set up the PyTorch with the packages shared in the link below:

Thanks.

15

Thanks for your reply @AastaLLL . I installed torch using the pip install torch --index-url jp6/cu126 index command. After that, when I tried to run inference, I encountered import errors for the transformers, av, torchvision, numpydantic, and pytorch3d packages. I was able to resolve most of them by installing each package individually using pip install <package_name> --index-url https://pypi.jetson-ai-lab.io/jp6/cu126 . However, I’m unable to install pytorch3d using the same method. I’ve attached a screenshot for your reference.

image
image1896×904 72.2 KB