maybe more of a question for @dusty_nv
i downloaded the jetson-containers for ML and ROS. I noticed they all came packaged separately. How does one build an app that utilizes multiple containers at once? I’m not super familiar with docker, sorry
Hi @VictorB, you would typically use one of them as your base container, and then create another container/dockerfile from that base container that includes the other contents that you want. You could do this a couple ways:
-
Let’s say you wanted to use the ros container, but have it include the l4t-ml container stuff too. The easiest way would be to change the base of the ros container to the l4t-ml container, and re-build it. You can set the base container of the ros containers here: https://github.com/dusty-nv/jetson-containers/blob/e6047853df66260bf567e820c24747bcdceebb4a/scripts/docker_build_ros.sh#L5
-
Create your own Dockerfile that uses l4t-ml as base, and copy in the commands from another Dockerfile (i.e. the ros dockerfiles) that install the packages you want/need. You can use the containers I provide in the jetson-containers repo as the base of your own container, and install any extra stuff in your own Dockerfile.
@Dusty_nv understood, thank you very much for the help! Really appreciate the good work you folks put into this project
We actually ended up finding another solution, which was to install Tensorflow in the host OS and only use the ros container. Seems to be working