Training Object Detection from Simulation in Docker

I am looking to follow this tutorial on an aws instance:
https://docs.nvidia.com/isaac/isaac/doc/tutorials/training_in_docker.html

Is this possible?

Yes, I believe it is possible to run it on an EC2 instance with a T4 GPU and using the NICE DCV to remote access the desktop.

Steps:

  1. Create an Ubuntu 18.04 (desktop version) EC2 instance with T4 GPU
  2. Install the NICE DCV server for remote access
  3. Install the other necessary drivers and Docker
  4. Remote access the EC2 instance using NICE DCV
  5. Run the tutorial on the instance like it is a normal desktop.

Would this be the correct tutorial to follow:

https://docs.omniverse.nvidia.com/robotics/text/Robotics_QuickStart.html#kit-remote

Sorry, the tutorial I was referring to was the one on your first post:
https://docs.nvidia.com/isaac/isaac/doc/tutorials/training_in_docker.html

The Omniverse Kit Remote Client is not necessary if you are accessing the EC2 instance using NICE DCV. The Omniverse Kit Remote Client is mainly used if you are deploying Isaac Sim on a headless instance.