[Unitree Go2] No CUDA installed on Nvidia Orin Nano 8 GB in Extension pack

Software
1

Hello,

we bought the Unitree Go2 EDU Version with the extension pack/Nvidia Orin Nano 8 GB. We want to use the robot in a lecture and want to run small neural networks like YOLO8n in an ROS 2 Python Node. This already works, but it can’t access the GPU because cuda isn’t installed.

Are there instructions on how to reflash the Nvidia Orin Nano in the extension pack using, e.g. the Nvidia SDK Manager? Do we have to unmount the Extension Board in order to access a specific USB-C port to flash it? As far as I know it isn’t possible to install cuda directly on an Nvidia Orin Nano.

It seems that the Nvidia Board in the Extension pack wasn’t (properly) flashed by mybotshop.

I would appreciate your suggestions on how to proceed.

Kind regards

2

Dear @roco-kit

MyBotShop is not responsible for setting up the Nvidia board that comes with the Go2 EDU, as it is preconfigured by Unitree. For installing CUDA, I recommend reflashing the image that we will provide containing CUDA on the Nvidia board. Please let us know which Go2 EDU version you have is it 100TOPS or 40TOPS?

3

Thanks for your support, we have the 40TOPS version.