Isaac ROS gpu activation in Docker

Hi,

I would like to try nvblox following this website:
https://nvidia-isaac-ros.github.io/concepts/scene_reconstruction/nvblox/tutorials/tutorial_realsense.html

I am using Jetson Orin Nano developer kit 8GB with Jetpack 6.0 and Intel Realsense D455.

When I tried the RealSense example, it ran smoothly. However, when I attempted the reconstruction with people or other dynamic scene elements, I noticed a display delay in the small window and a significant reconstruction delay, even though everything launched successfully.

I suspect the GPU was not working because when I tried these tasks in Docker, I checked nvidia-smi and saw something like this:

No processes were running on the GPU. Is there a way to check if the GPU is working or to enable it manually?

Any ideas/suggestions/help would be super helpful! Thanks!

Hi @AustinCTH

Thank you for your post.

I recommend using jetson-stats for improved output on NVIDIA Jetson devices. Additionally, there is an Isaac ROS package available to monitor diagnostic messages.

Best,
Raffaello

1 Like

Thanks a lot @Raffaello . I just realize that I have turned off the fan by watching jtop, and my Jetson performed much better after I turned the fan on again.

1 Like

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.