Tuesday, November 9, 2010

Adding value with machine vision

“Lean” has been the way many manufacturing companies want to do things for quite a few years, and there are many “Lean” gurus running around telling companies what they need to do. A central tenet of “lean” is the elimination of waste, and waste is defined as anything that doesn’t take the product closer to its finished state.

With a definition like this, it’s easy to see why the disciples of Lean, (and I count myself among their number,) frown on inspection: “Just sorting good from bad,” they say, “Better to not make any bad ones.” And who can argue?

The problem is that resolving problems and reducing process variation takes data, lots of it, and good process data is often hard to find.

Machine vision is the perfect tool for gathering process statistics, whether a dimension, a shade, or some other attribute. It doesn’t sleep, never goes on break, and is rarely distracted by the pretty girl from Accounts Receivable, so why don’t we make more use of it for data capture?

Endre Toth of Vision Components has some thoughts on this subject, which he shared with Test & Measurement World in “Vision, inspection data improve processes,” (October 2010.) It’s not an exhaustive analysis but it should get you thinking.

No comments: