ROS and Conda

[Cody-Vu]

Hello everyone,
I have a quick question about this wrapper. So I have a real time detection transformer which detects objects and uses conda environment. Now I would like to publish the positions of the object by initiating a ros node in the transformer. On the other side I have a normal python script for path planning and controlling a robot using ROS. It would be nice if the my python script had access to the output of the object detector.

Would this wrapper work for my use case? And if yes would I need to do everything in the conda environment, or is it possible to publish the positions in the conda environment and outside of it other ROS-nodes would be able to subscribe and have access to the positions? Maybe there is also a better option to handle my task and I'd be glad to know.

Thank you very much in advance and I'd be grateful if someone could clarify!

[traversaro]

On the other side I have a normal python script for path planning and controlling a robot using ROS.

I am not sure what you mean by "normal" python script in this context. You mean a Python script that is using ROS and Python installed via apt?

In general I would expect the same distro in robostack and the same ROS distro installed by apt to be able to communicate over topics. Communication across different distros is instead known to be problematic.

[Cody-Vu]

Thanks for your quick response. Yes with normal script I meant a script using ROS and Python installed by apt.

In general I would expect the same distro in robostack and the same ROS distro installed by apt to be able to communicate over topics. Communication across different distros is instead known to be problematic.

So that I got it right: If I use robostack you recommend me to run everything in the conda environment? Do you know an easy option to access the positions of the object detector from my python script?

[traversaro]

I think it is slightly easier to have everything in the conda environment, as in general we do not support to source the apt installation of ROS, and then activate the environment, but that depends a lot on the other constraints of your problem.

Do you know an easy option to access the positions of the object detector from my python script?

This is a really general question that is difficult to answer. If the object detector publishes the positions on a ROS 2 topic, you can read them via rclpy.