diff_drive_controller shall accept velocity feedback from motors

[roncapat]

At a certain point in ROS1 implementation, estimate_velocity_from_position parameter switch was inserted, and odometry could also be computed by integration of velocity values coming directly from motors.

In ROS2, this in not an option. Although somewhere it is written that the controller works "through a velocity interface" this is true only for the input chain, not the feedback one.

I propose a port of the feature also in ROS2 , and at least specify clearly what are the required Hardware Interfaces (in/out) for this controller in the documentation.

[bmagyar]

Yes! I'm quite sad that we never got to port more of the diff drive functionality over to ROS2 as we are still fairly busy with the framework itself. I'd be very happy to review some PRs for this though!

[destogl]

@roncapat do you have capacity to do this?

@joshnewans do you maybe have interest in this topic?

[roncapat]

If you are interested, I can try to develop and push a PR whenever I find some spare time.

[joshnewans]

Position feedback is sufficient for my current needs but I agree that velocity feedback would be good.
I think @roncapat would be better positioned if he has a need for it, but otherwise I'd consider taking a look down the track once I find time.

[roncapat]

I searched again the sources of the implementation I mentioned in the opening of this issue. From mobile, didn't found yet the code I guarantee to have seen at some time in some branch of ros_controllers official repo, but I remember it was something very similar to what can be seen here:
clearpathrobotics/ros_controllers#10

Starting from this code, I think that in the next week I can open a draft PR, or maybe even already test it in my company on our retrofitted Seekur Jr Rover.

[roncapat]

I just posted a PR. Should work, I need that someone tests it with position_feedback: true just to acknowledge that there are no regressions on old code with position feedback. The implementation is mostly straightforward and easy to validate preliminarily by source code analysis.

[roncapat]

@bmagyar @destogl could you unlock the checks on the PR #260?

[roncapat]

Closing since #260 got merged.