Component parser: Get mimic information from URDF

[christophfroehlich]

I started implementing the parsing of information about mimic joints from the URDF (instead of duplicating information in the ros2_control tag).
This will fix #652:

After a discussion with @destogl it seems we converged on a solution very close to the above.

Overview: Using a mimic joint should work out-of-the-box, on by default. Users who want to opt-in can do so in the configuration of their hardware component. The mimic property should be defined in the wider URDF section, not under the <ros2_control> tag.

In the example with the setup below, ros2_control should do the check on each joint to figure out if it is a mimic joint or not. This could be part of JointInfo (will need to reach up to the full urdf to find it) ((implementation detail))

<joint name="right_finger_joint" type="prismatic">
  <axis xyz="0 1 0"/>
  <origin xyz="0.0 -0.48 1" rpy="0.0 0.0 0.0"/>
  <parent link="base"/>
  <child link="finger_right"/>
  <limit effort="1000.0" lower="0" upper="0.38" velocity="10"/>
</joint>
<joint name="left_finger_joint" type="prismatic">
  <mimic joint="right_finger_joint" multiplier="1" offset="0"/>
  <axis xyz="0 1 0"/>
  <origin xyz="0.0 0.48 1" rpy="0.0 0.0 3.1415926535"/>
  <parent link="base"/>
  <child link="finger_left"/>
  <limit effort="1000.0" lower="0" upper="0.38" velocity="10"/>
</joint>
<ros2_control>
  <joint name="right_finger_joint">
    <command_interface name="effort"/>
    <state_interface name="position"/>
    <state_interface name="velocity"/>
    <state_interface name="effort"/>
  </joint>
  <joint name="left_finger_joint">
    <command_interface name="position"/>
    <state_interface name="position"/>
    <state_interface name="velocity"/>
    <state_interface name="effort"/>
  </joint>
</ros2_control>

If one wants to disable mimic behaviour for some reason, they can still turn this functionality off by adding mimic="false" to the joint like so

  <joint name="left_finger_joint" mimic="false">
    <command_interface name="position"/>
    <state_interface name="position"/>
    <state_interface name="velocity"/>
    <state_interface name="effort"/>
  </joint>

Originally posted by @bmagyar in #652 (comment)

Decisions in WG meeting on 2024 01 03:

• mimic="false" should be for opt-out; missing or mimic="true" has the same effect
• component_parser throws an error if an invalid URDF is given. Additionally, I added a check if a joint defined in the ros2_control tag is available in the URDF and throw an error otherwise. Is this ok, or are there any use-cases for an incomplete URDF?
• As a consequence of the above point: The tests failed, because there are some interfaces like "configuration/max_tcp_jerk" defined as joint. IMHO we should use the semantic information of the type and changed them to be of gpio type.

migration notes

I moved existing migration notes directly below the ros2_control heading and added migration notes for this change. Or should we put them at top-level? (can be done in a different PR).

[codecov]

Codecov Report

Attention: Patch coverage is 89.23077% with 7 lines in your changes are missing coverage. Please review.

Project coverage is 76.14%. Comparing base (35bb5f7) to head (3a1c1cd).
Report is 15 commits behind head on master.

❗ Current head 3a1c1cd differs from pull request most recent head 931f108. Consider uploading reports for the commit 931f108 to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1256      +/-   ##
==========================================
+ Coverage   75.51%   76.14%   +0.63%     
==========================================
  Files          41       41              
  Lines        3328     3396      +68     
  Branches     1921     1957      +36     
==========================================
+ Hits         2513     2586      +73     
+ Misses        421      420       -1     
+ Partials      394      390       -4     

Flag	Coverage Δ
unittests	76.14% <89.23%> (+0.63%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files	Coverage Δ
...terface/include/mock_components/generic_system.hpp	100.00% <ø> (ø)
...e_interface/src/mock_components/generic_system.cpp	79.18% <100.00%> (-1.17%)	⬇️
hardware_interface/src/component_parser.cpp	92.83% <88.88%> (-0.42%)	⬇️

... and 5 files with indirect coverage changes

[saikishor]

I'll continue my review for the other files later

[mergify]

This pull request is in conflict. Could you fix it @christophfroehlich?

[saikishor]

Rest of the changes, look fine to me

[saikishor]

The changes look good to me.
Just few minor suggestions :)

[mergify]

This pull request is in conflict. Could you fix it @christophfroehlich?

[mergify]

This pull request is in conflict. Could you fix it @christophfroehlich?

[mergify]

This pull request is in conflict. Could you fix it @christophfroehlich?

[saikishor]

Niceee!
Thanks for fixing it. Can we get this merged soon? I have 2 more PRs waiting for these changes.

[destogl]

Can you open follow up on this by removing it add ading migration notes to ROS-J. Even better make 2 PR from it, so we can merge already migration notes, but then later (end of april) delete the code.

[christophfroehlich]

#1463

[destogl]

This should be also marked in the migration notes

[christophfroehlich]

done. I added a screenshot in the PR summary on top.

[christophfroehlich]

@Bence @destogl Let's merge this and then ros-controls/control.ros.org#262 to put the migration notes on the top level of the docs.

[bmagyar]

Say no more!