Wednesday, February 25, 2009

Does machine vision have any place in the “lean” factory?

If you take the view that machine vision just sorts good parts from bad, then perhaps the answer should be “no.” I would argue though that vision is central to the core “waste elimination” tenet of “lean,” for two reasons:

First, machine vision is an excellent way to error-proof (“poka-yoke,” in the lean lexicon,) manufacturing operations. Vision systems can easily verify that a part has been included in an assembly, as well as checking for correct orientation and detecting missed operations (by looking for specific or unique features.) But more than this, they can be quickly and easily reconfigured, either to accommodate a design change, or to be redeployed elsewhere in a manufacturing plant.

Second, machine vision is an excellent tool for monitoring trends, which is where it can help reduce process variability. For example, in gauging applications a vision system can reliably generate streams of statistical data that will show when a process is drifting out of control. With just a little engineering effort it’s even possible to set up a feedback loop to act on the information from the vision system. Machine vision is even more useful when the inspection characteristics are difficult for the human eye. A great example of this is color: it’s very difficult for a human to detect a gradual change in shade.

So to anyone who claims that machine vision is unnecessary in a “lean” factory, I say this: see how lean you can get by relying on human eyes instead.

No comments: