Jetson Newbie in search of a development environment/process

New to containers and frankly I am finding them to be the biggest impediment to my progress. I don’t understand, yet, how to deal with every time I run a container I’m in a new environment, none of my preexisting tools are there, apart from whatever shared volumes I set up in the docker run command. There’s tons of AI and dev tool etc. demos floating around but they’re all in docker containers and I haven’t figured out yet how to either a) strip what I want out of the containers or b) merge containers.

I would like to have an actual source level debugger a la vscode, Pycharm, etc. (I was using Wingware before and loved it), and sew together demo code like the “Jetson AI Fundamentals - S1E2 - Hello Camera” but I can’t get past “this container is not your environment”. I understand it may be unavoidable that I have to learn all about containers before I can write “hello world” but now I’m in “Hell World”. So I’ll ask some specific questions and hopefully people will be kind enough to chime in with some advice.

  1. Are containers absolutely necessary? MUST I learn all about Docker to continue with Jetson? (it sure seems that way now)
  2. How do people debug? AFAIK VSCode requires minor surgery to let you debug python in a container. I’ve started playing with Pycharm but haven’t tried to deal with containers yet.
  3. What is YOUR general development environment? i.e. do you have your own custom container? What do YOU do when given a container with some code or other software (e.g. ONNX Runtime) that you want but you certainly don’t want to just work in that container without all your preexisting dev tools?
  4. How does NVIDIA envision newbies to go from prepackaged " Jetson AI Fundamentals - S1E2 - Hello Camera" to full fledged development?
    I hope this makes sense and someone can give me some advice, I’m pretty frustrated how long it’s taken to go from running the simple " Jetson AI Fundamentals - S1E2 - Hello Camera" to writing my own code in a proper debugging environment. I feel Docker is the main obstacle since I never had this much trouble building things with the Raspi, Beagleboard, CHIP, etc. Thanks in advance, sorry if I’m whinging.


1. No. You can do everything outside the container directly.
It is used for user to have the same environment that a tutorial based on.

2. Yes. Some of our users use PyCharm.

3. If you want an customized environment, you can build a docker on your own.

4. Maybe you can check our jetson-inference tutorial.
It doesn’t require a docker environment and can lead you to train a model and deploy it on Jetson.