What should I compare when our UR5e sleeve grip changes between batches?

We retained an earlier sleeve and three from the troublesome batch. Maintenance has the finger drawing. I've stopped treating the tray label as evidence that everything else stayed the same.

Do the clips show the sleeve moving during closure, or already sitting crooked beforehand?

One side moves as the fingers close. It looks upright before that, but the front view hides the opposite contact. I have marked that uncertainty in the comparison.

Have inspection compare the relevant sleeve dimensions and contact faces against the current drawing alongside maintenance's isolated assembly check; otherwise a difference between batches could be blamed on the tool without checking the parts.

Who is keeping the job unavailable while that is checked, so the covering shift doesn't inherit an instruction to get it going with another small correction?

Our shift lead owns that, Ada. The job is held for the check. The later batch is also separated, but not yet declared bad; inspection hasn't reported back.

And are the saved edits tied to those batches, or just numbered versions?

Numbered versions with change times. Two can be related to the operator's batch notes; the earliest one cannot. I won't invent a reason for it now.

That missing reason belongs in the review too, because restoring an earlier version would still be a choice someone has to justify against the physical setup.

Can inspection check several current sleeves rather than only the one with the most obvious tilt, while retaining that specific example for comparison with the observed contact?

They did. Two later-batch sleeves are outside the specified roundness requirement at the gripping region; the others checked so far meet it. Maintenance found the finger and pad assembly consistent with its drawing.

Does that explain the exact sleeve in the clip?

Yes, its identity was retained and it is one of the two. Quality has held the batch for sorting. We still need the controls review and supervised transfer check on accepted sleeves before releasing the job.

Will the incoming check cover that gripping region now, rather than leaving the first robot miss to discover the next questionable batch?