EuRoC Forum

Hi,

I've noticed a problem with the TF frames of the holes w.r.t. the 3D model of the workpiece.

An example is showed in the image below.

Hi Fabjan,
I think you are facing the same problem exposed by AIS in this topic:
20112706nx57588/challenge-1-f3/-stereo_camera_link-and-left_camera_optical_frame-t121.html

I hope we will receive news about it soon.

Best.
Stefano

Hi michieletto,

thanks for the answer, the problem is not in the stereo frames but in the holes TF frames which are not aligned with the holes in the model.

another problem in the holes TF frames is the position accuracy. An example, as you can see the h18.position.z value is 0.0995 m and not 0.1 m, but we have to detect the holes position with 0.3mm of accuracy. So we are already over the error threshold for the 10 points because the hole detection can be only in the workpiece surface (of course without an empirical correction).
<node args="0.0753 -0.09 0.0995 0.0 0.0 0.0 1.0 workpiece_link h18 100 " name="stp18" pkg="tf" type="static_transform_publisher"/>

We can also confirm that with the latest update, the holes of the workpiece in Gazebo, no longer match the frames published in TF.

Dear all,

we found and fixed the problem. Sorry for the inconvenience.
20112706nx57588/challenge-1-f3/updates-for-challenge-1-t8.html

Best