Is it possible/allowable to clone the appropriate docker containers and run these on a Ubuntu machine separate from the actual controller? I’d like to have the simulated robot running on a machine at home to make it easier to develop against.
Hi Robert. Thanks for the response. Correct me if I’m wrong but the Hub version doesn’t allow us to set the ROS_MASTER_URI on a local machine and then interact with the simulation’s ROS/Hal.
Our specific desire is to create a local ROS node running a machine vision service and have the simulation react to data coming from this node. We’d like to be able to develop this system without access to the ZA-6 controller.
To this end, I spent today digging into how the two docker containers on the controller are spun up. I can see, for example, that ros-dist-launch eventually uses the pp_ros_launch python package to craft and do a “docker outside of docker” start up of the ros-None-ui container. Is it okay to go ahead and manually do this on another Linux box? If so, any obvious gotchas to watch out for?
Thanks for the additional info, I feel like I understand the need here - Enabling a workflow to run robot simulation on a non-controller machine in a more development-friendly environment. This isn’t something we currently officially support at the moment. I really like this idea and I’ve created a ticket for us internally to track and prioritize.
Hi Brandon, thanks for adding this to your internal ticketing system/roadmap.
At this end, we managed to get it working by understanding how the simulator container gets launched on the controller, recreating the docker file to launch this, and migrating it all over to another machine. This thread (Running a Modified Copy of the Docker Image) will be of help to anyone trying to do similarly.