[PID][pid_controller] "i_clamp" is active even when "antiwindup" is set to false.

[ViktorCVS]

Describe the bug "i_clamp_max" and "i_clamp_min" is limiting the "i_error" even when "antiwindup" is set to false. The description of "i_clamp_max" and "i_clamp_min" given the understand that they only limit the "i_error" when "antiwindup" is set to true.

To Reproduce Steps to reproduce the behavior:

1. Set a PID controller with "antiwindup" set to false
2. Set "i_clamp_max" and "i_clamp_min" to 0.0*
3. Set p and i different from 0.0
4. Execute the simulation** with the controller and send a reference
5. See the error - in this case, the 'i' pid parameter doesn't do anything, because "i_clamp_max" and "i_clamp_min" is capping the "i_error" to 0.0.

• Even when I don't set them, due to the dafault value, they are set to 0.0 ** I faced this issue while on simulation

Expected behavior I expected that if "antiwindup" is set to false, "i_clamp_max" and "i_clamp_min" won't cap the "i_error" and the "i" parameter would have effect on the control.

Screenshots

Environment (please complete the following information):

• OS: Ubuntu 22.04.4 LTS
• Version: Iron
• ros_demo_example_1 modified to velocity control

Additional context In screenshot, is visible that besides the "i_error" is increasing, the output is constant. The reference of velocity, 0.1, never get achieved. In RVIZ, the simulation is coherent with the data. I made new tests, setting "antiwindup" to false and "i_clamp_max" and "i_clamp_min" to a high value, like +-100000 and it worked and "i_error" isn't limited.

Output can be read by topic: /gains/joint1/pid_state

[christophfroehlich]

I had a quick look, it should be considered with the PID ros2_controller. And this uses the pid_ros from control_toolbox package, where we have tests for that.

Only to rule out the simple mistakes: Have you checked if the correct parameter is set? I has to be gains.<joint_name>.antiwindup for every joint. (for example, use rqt_dynamic_reconfigure).

[ViktorCVS]

Thanks for the answer!

Here the full code block from controllers.yaml

besides this, "antiwindup" parameter is default set to "false"

[ViktorCVS]

I'll close it because it is a expected behaviour of the PID controller from control_toolbox pid code.