best practice about creating new node from existing one

edit

moveit_servo holds indeed panda specific values.

Best practice would be to have a generic upstream package , that way you can git clone it and add your robot specific files. Then you can just do a git pull to get the latest changes. However this is in my opinion not realistic for moveit_servo because there are hardware parameters required. moveit_servo without panda reference wouldn't be a working simulation, or it would be over abstracted.

in this case (moveit2 is rolling, and this specific package is active on git) I'd keep the panda stuff, and add your robot configs in new files (i.e: "myrobot_simulated_config.yaml"). Then when you pull an from git you can easily compare diff between your local and remote. (for release/distribution clean it up and remove panda references)

There's nothing about Servo that's specific to the Panda. It should work well for any serial manipulator. There are some config files in the repo that are based on the Panda -- these are for testing and tutorial purposes only.

If you want to use your own unique config files, you just need to make a new launch file to point toward the right package. These are the lines to change (for example):

https://github.com/ros-planning/movei...

https://github.com/ros-planning/movei...

https://github.com/ros-planning/movei...

https://github.com/ros-planning/movei...

It sounds like your other idea is basically to fork the Servo package then make modifications of the config files. That would work but then you'll be building a lot of code from source that you don't have to.