PID controller and diff drive controller chaining

[Juliaj]

Address #568. This is to add a new example_16 based on the use case outlined in the controller manager doc section Motivation, Purpose and Use. It uses example_2 as a base and adds a chained 'diff_drive_controller' and two PID controllers with following flow.

diff_drive_controller     → pid_controllers      → hardware
     (velocities)           (error control)       (motors) 

This requires following changes from ros2_controllers:

• Fix handles for following controllers of diff_drive_controller ros2_controllers#1513
• [pid_controller] Fix logic for feedforward_mode with single reference interface ros2_controllers#1520

New Examples

If you propose adding a new example, make sure that your new example has:

• The correct folder structure (described in the main README.md)
• Example has doc/README.rst with description
• Detailed commands how to run an example
• Output examples of CLI commands
• doc/index.rst is updated.

[christophfroehlich]

I think you found a bug, I proposed a fix in ros-controls/ros2_controllers#1513
I also changed your configuration a bit, let met know what you think.

[christophfroehlich]

Thanks for your submission! Some thoughts:

• Please cleanup the docs to focus on the controller chain (e.g., remove the MockHardware part)
• Update the test_diffbot_launch.py to check for all controllers
• I suggest that you add some dynamics to the hardware interface, a first-order or second-order low pass, so that the PID controller actually makes sense.

When the following stuff is fixed/merged we should include to the docs (maybe in follow-up PRs)

• view_controller_chains output Fix view_controller_chains ros2_control#2026
• add plots of pal_statistics to show the PID controller action Integrate pal_statistics for introspection of controllers, hardware components and more ros2_control#1918

[Juliaj]

I think you found a bug, I proposed a fix in ros-controls/ros2_controllers#1513 I also changed your configuration a bit, let met know what you think.

Thank you so much for responding so quickly. The configuration changes are really nice, love them. I tested the fix you made, the activation is working now. I'll continue the rest of work and include the suggestions that you made. Left some comments on your PR ros-controls/ros2_controllers#1513.

[christophfroehlich]

see #1528 for a proper fix of the interfaces. could you have a look there and review please?

[christophfroehlich]

The demo (and the chaining itself) is working great now, thanks for the effort! I already have added the doc page to the TOC, some other comments below.

I have tuned the PIDs a bit to have a faster step response. what do you think about adding the plotjuggler config and a screenshot to the docs? Feel free to record any useful time series.
pj.xml.txt

[Juliaj]

I have tuned the PIDs a bit to have a faster step response.

This is truly awesome. I must admit that while I’ve known about PID controllers from textbooks, this is the first time I’ve actually had the chance to play with one.

what do you think about adding the plotjuggler config and a screenshot to the docs? Feel free to record any useful time series.

Thank you so much for pointing me to the plotjuggler tool, really love it ❤️. I added a few python scripts with hope to simplify some of the steps. Let me know your thoughts on these.

I also has an issue with plotjuggler which I'm not able to get data for pid_controller command interfaces, i.e. there is no plot for command_interface.pid_controller_left_wheel_joint/left_wheel_joint/velocity. Is there additional step that I need to do to accomplish this ? Once I get a proper diagram, will add to the documentation.

[christophfroehlich]

Thanks, this looks great now.
With ros-controls/ros2_control#2050 the command reference interfaces are reported here, can you confirm this?

[Juliaj]

Thanks, this looks great now. With ros-controls/ros2_control#2050 the command reference interfaces are reported here, can you confirm this?

That's it. Added the diagram!

[christophfroehlich]

Two new comments, and still some older open conversations, can you have a look please?

[christophfroehlich]

LGTM. Just let's wait for the next release of ros2_control and ros2_controllers before we merge this.

[christophfroehlich]

The changes on ros2_controllers got synced to ros2-testing on rolling today, so we can merge that now