Our physical plate judgements are checked and the raised-support experiment is out of the ordinary reference set, but a bright band still changes in the normal seated view when the workshop door opens; I want to understand that lighting change before someone adds another scratch arrow to our FR10 examples.
Hold the identified plate and camera setup constant while comparing the door conditions, with the actual setup recorded. Ask the vision engineer whether changing ambient light is entering the scene before tuning around those different images.
And include the real rejected plate in that comparison, Ravi, not only the accepted one whose band is annoying you; removing the band is not the whole inspection job.
Both included, Naomi; the band changes with the door while the inspected groove stays at its physical location, and the engineer has reproduced that under the recorded setup rather than guessing from my two favourite photographs.
Compare a controlled lighting arrangement that works with the ordinary door use. Don't make the operator responsible for remembering to keep a workshop door shut if that is not how the workshop actually runs.
A defined local screening arrangement around the camera view, with the loader's access and the robot presentation included; I have asked for that whole arrangement assessed, not an instruction to drape something over the bench.
Included in the revised study price, Naomi; reviewed screening is fitted for the bench comparison and the door-dependent band is now controlled, with our actual groove still visible through repeated presentations.
Include representative accepted finish variation and another rejected feature location with quality. That avoids selecting an arrangement which happens to favour only the first two retained plates.
Thanks for recording the door condition and the physical-reference result separately. Has that broader comparison finished, or is the improved scene still a candidate rather than a completed inspection?