简体中文界面。帖子、指南及政策正文保留原文;未对原文作自动翻译。

Why are we tuning the UR5e cover inspection around unverified scratches?

OwenBrown0921 · 2026年6月10日 11:07 UTC

回复讨论
OW
OwenBrown0921
Our planned camera station has bright streaks labelled as scratches, although the streaks move with small presentation changes. I am uneasy about tuning against those examples before commissioning. How would you separate physical appearance judgements from whatever the camera happens to highlight?

15 条回复

EM
EmmaArcher0376

I'd take the labelled examples back to the physical covers first, Owen. If nobody can point to the claimed damage under the agreed inspection conditions, the label needs to remain a question.

22
OW
OwenBrown0921

I can identify two covers from the images. One was accepted manually, but it has a scratch label in the camera folder. The image label came from its bright streak.

12
CH
ChloeBaker0483

Keep the accepted cover and that image together. It's a useful example of the labelling mistake, provided it stays out of the confirmed-reject set.

10
HA
HazelArcher0397

We had a photo everyone agreed looked dreadful, then the actual piece was fine. The image was on a slide with heavy contrast. Is your team judging original captures or compressed copies?

9
OW
OwenBrown0921

Mixed copies, including screenshots. I have asked for the originals and their setup notes. The physical acceptance record is available for the identified cover.

9
HA
HarishBell0663

Who makes the call on the other cover? Two people can agree the first is good and still spend an afternoon inventing different personal standards for the second.

2
OW
OwenBrown0921

Quality lead is responsible. The second cover is now awaiting her assessment. I have not treated it as a reject just because its photo looks more dramatic.

4
EM
EmmaArcher0376

When she has decided, compare fresh captures of those known covers with the settings recorded. I'd include taking them out and putting them back, because that was part of the uncertainty you described.

19
HA
HazelArcher0397

Owen, did you recover the original images? Curious whether the screenshots were amplifying anything, as they did in our earlier case.

23
OW
OwenBrown0921

Originals recovered. The streak is still there, so screenshots were not its whole explanation. Quality rejected the second cover for a separate edge scratch and recorded the location.

4
CH
ChloeBaker0483

Show that edge in the planned camera view too. A visible harmless streak shouldn't distract everyone from whether the real reject feature can be inspected.

19
OW
OwenBrown0921

The edge scratch is sometimes obscured after reseating. We are reviewing the holder and light positions before another capture set. Thanks Chloe, that was the question the old folder never answered.

5
HA
HarishBell0663

Who will explain inconclusive images to the operators in the handover? The people collecting rejects shouldn't have to decide whether your reference picture happened to be the lucky angle.

9
OW
OwenBrown0921

Quality and the trainer are defining that route. Inspection acceptance remains open while the imaging comparison continues; I am keeping those limits in the handover draft.

22
HA
HazelArcher0397

Please return with the holder result. I like hearing when the answer is a better way of presenting the part, even if a new camera would make a more exciting purchase request.

14

参与讨论

欢迎来到 Application Robot

所有人都可以阅读论坛。登录或注册后即可发起讨论、回复或上传照片。

忘记密码?

注册账号即表示你同意我们的 使用条款 社区准则。请阅读我们的 隐私政策 ,了解我们如何处理你的信息。