Suggestions on development cycle

[aditya1709]

Might be a silly question, suppose I wanted to change the mission, add assets or even change the robot configuration I'd have to change the server side implementation. In the current workflow we have to export it as an extension to IsaacSim and this does not seem to be a efficient way if we are iterating on the scene/robot configuration etc. What is the preferred way to iterate on the server side without having to export it to .zip and then add as extension to test?

Additionally if the client was on an actual hardware how would this workflow change?

[liorbenhorin]

@aditya1709 Hi,
From an architecture - This repo treats Isaac Sim side as the client, and the the container at /isaac-zmq-server as the server.

The repo build.sh is taking care of copying the build artifacts and placing them back into /exts of the repo root.
This is designed to make the development process easier - you should only set the path to /exts once in isaac sim, and keep developing normally in the repo.

• If all the changes you make are in the python files, you will enjoy hot-reloading so you can easily change mission parameters, assets, etc etc. (no need for a build)
• If you want to make a change the to protbuf messages, or the c++ zmq node, then you have to re-build, and also I recommend restarting Isaac Sim.

If the server is not running on the same host, you should provide the server ip on Isaac Sim side - here:

IsaacSimZMQ/exts/isaacsim.zmq.bridge.examples/isaacsim/zmq/bridge/examples/example_missions.py

Line 44 in 72033fe

self.server_ip = server_ip

[aditya1709]

Thank you and did you have some guidance on how one could integrate ROS stacks like Nav2 etc without having to use ROS bridge?
Ideally if I build out the ZMQ bridge to interface with a model (VLM, VLA etc) I'd prefer using the same bridge with controllers like Nav2 as well instead of using ROS bridge that most resources talk about. Any thoughts on how to integrate SLAM/NAV packages with ZMQ? Should we write a ZMQ/ROS bridge interfaces in that sense and actually have a ROS bridge as well?

[liorbenhorin]

Hi @aditya1709 sorry, must have missed this comment,
This is a matter of preference and design choices / simplicity / development velocity etc.
Sometimes it can make sense combining different protocols/applications, sacrificing performance but validating some ideas.
Optimization usually comes later.
So for your question, if it will be sub-optimal - but will help you make progress - do it :)