Missing detail - static tf can not be disappear from rviz2

peterborkuti · December 23, 2023, 10:17am

Hi,
See the screenshot and the selected text:

Frame Timeout: The length of time, in seconds, before a frame that has not been updated is considered dead. For 1/3 of this time the frame will appear correct, for the second 1/3rd it will fade to gray, and then it will fade out completely

So far in the course only static tf was used. When I started playing the mentioned setting and I stopped publishing map->table1 static frame, I noticed, that frames in RVIZ2 do not fade out.

I found this explanation:

github.com/ros/geometry2

Add possibility to delete static TF frame.

opened 05:09PM - 04 Mar 19 UTC

peci1

With tf2, there's no possibility to remove already published static TF frames. I… think it'd make sense to have such possibility. There's a relevant ARO question: https://answers.ros.org/question/311864/is-it-possible-to-delete-a-static-tfs/ . My use-case: a sensor is removed from the robot, I regenerate the URDF, restart robot_status_publisher, but RViz and all other nodes are still showing the removed frames until they're restarted. I'm not sure about the ideal implementation of such a thing. My idea: static_transform_broadcaster could get a `removeFrame()` or `clear()` method, remove the frame from its `net_message_` cache, and publish to `/tf_static`. The listeners could notice that they received fewer frames from this specific publisher, and seek for the ones that should be deleted from their caches. However, that'd need storing the publisher id with each transform in the cache, which is probably not a good idea. So maybe there could be a special value of `TfMessage` that'd instruct listeners to remove the frame from their caches. `NaN`s can't be this value, because `TransformListener` ignores all messages with `NaN`s and changing this behavior would break a lot of code... Infinity also doesn't seem very good... Any ideas? Maybe misusing timestamps by storing the maximum uint value?

Which beautifully explains, why will not static tf will ever fade out.

Because learner can be confused the fact that some of RVIZ2 settings “does not work” for static tf, I think, this explanation can be put into the material.

Thank you
Péter

bayodesegun · December 27, 2023, 9:59am

Hi @peterborkuti

Thanks for the feedback! Our courses team will review that and may include it in the notes.

duckfrost2 · December 27, 2023, 2:28pm

Hi,

We added a small line stating this fact with static transforms in the humble version.

peterborkuti · December 27, 2023, 3:51pm

Thank you. I enjoyed the course very much.

system · January 1, 2024, 3:51pm

This topic was automatically closed 5 days after the last reply. New replies are no longer allowed.