doc: Added explanation of preceding/following controllers

[peci1]

Resolves #2189 .

This PR adds explanation of what preceding/following controller means.

I also broke the rule and attached an irrelevant fix of grammar in the paragraph directly preceding the paragraph I added. I hope it's not a big no no :)

[christophfroehlich]

not sure about the RST syntax with '', let's see what happens

[bmagyar]

Thank you!

[peci1]

I have one more unclear point, probably because I'm pretty new to ros2_control: Could "if A's output" be replaced with "if A's command interfaces"? Or are command interfaces somehow different from controller outputs? I may be a bit confused because for Hardware, command interfaces are inputs...

[christophfroehlich]

I have one more unclear point, probably because I'm pretty new to ros2_control: Could "if A's output" be replaced with "if A's command interfaces"? Or are command interfaces somehow different from controller outputs? I may be a bit confused because for Hardware, command interfaces are inputs...

There is no "output" in the implementation actually. A following controller or hardware component exports command interfaces, and (preceding) controllers are claiming it (as output). I.e., controller A is claiming the reference interfaces of controller B in this example. Should we also phrase it like this?

[peci1]

Hmm, then I'm probably missing a distinction between command interface and reference interface...

[saikishor]

Hmm, then I'm probably missing a distinction between command interface and reference interface...

The command interfaces are the outputs of your controller, where as the reference interfaces are the inputs to your controller so that they can produce outputs (which are commands).

For example, for a diff_drive_controller, the command interfaces would be the velocities of the wheels, where as the reference interfaces can be the base twist velocity. I hope this is clear for you :)

[peci1]

Hmm, still not there... It seems to me @christophfroehlich and @saikishor have conflicting views:

There is no "output" in the implementation actually.

vs

The command interfaces are the outputs of your controller

Also, this type of code (taken from diff_drive controller) doesn't help my understanding of the distinction:

reference_interfaces.push_back(
    hardware_interface::CommandInterface(

[christophfroehlich]

From the signal flow, sure, the controllers have an output. From the memory side they don't create an output, but they claim them.

[peci1]

Okay, I probably start getting this.

A precedes controller B if A's output is connected to B's reference interface.

Could it be

A precedes controller B if A claims B's reference interfaces.

?

Or is the formulation with outputs more clear?

[saikishor]

Okay, I probably start getting this.

A precedes controller B if A's output is connected to B's reference interface.

Could it be

A precedes controller B if A claims B's reference interfaces.

?

Or is the formulation with outputs more clear?

Yes, you are right!

[peci1]

Or maybe to get the best of both the precise and the intuitive wording:

A precedes controller B if A claims (connects its output to) B's reference interfaces (inputs).

[peci1]

Thanks, I've updated the wording.

[christophfroehlich]

Thanks for the discussion, now it looks fine!