add TF/Pose_V pub in DiffDrive

[doisyg]

Solves gazebosim/ros_gz#131
With this PR, there is only /tf@tf2_msgs/[email protected]_V to add to the args of the ros_ign_bridge parameter_bridge node to get similar TF publishing behavior as in Gazebo Classic DiffDrive plugin.
Once merged I can PR ros_ign_gazebo_demos if needed with the additional /tf@tf2_msgs/[email protected]_V.

I did the most straightforward implementation. Improvement to discuss:

• Variable naming: tfPub, tf_msg or pose_VPub and pose_V_msg ?
• Ign topic for frame_id -> child_frame_id transform: tf or something else ?
• TF/Pose_V always on, or enable/disable for instance through _sdf->HasElement("publish_tf") (or other name)

[chapulina]

Thanks for the PR! Did you consider using the PosePublisher system alongside the DiffDrive?

[doisyg]

I wasn't aware of the PosePublisher plugin but I guess that could work too!

[doisyg]

Though if it is the plugin that publishes directly the Pose_V / tf, we get a behavior/configuration more similar to the gazebo classic in my opinion.

[chapulina]

if it is the plugin that publishes directly the Pose_V / tf, we get a behavior/configuration more similar to the gazebo classic in my opinion.

Agreed, but in the long run, less duplication is easier to maintain. If we need to update something about how transforms are being published, right now we only need to update the PosePublisher and all users benefit from it. I'm inclined to suggest users migrating from Gazebo classic use the 2 plugins, instead of duplicating the functionality here.

[doisyg]

I gave it a try but I fail to understand how to configure PosePublisher to publish the transform from odom to the link of the robot.
What model/link would be odom ?

[doisyg]

Coming back on this, I cannot find a solution to get the tf published from frame_id to child_frame_id without this PR and by using the PosePublisher. Would you reconsider the merge ?

[chapulina]

What model/link would be odom ?

Good question. I can see now how this PR is needed. Besides addressing the minor comments below, do you mind also documenting the new topic in DiffDrive.hh? Thanks!

[chapulina]

Thanks for iterating, looks good to me. I just have one final question about the test.