Sunday, March 10, 2013

A vision system selection case study

There are some skilled salesmen in machine vision, and I often get brought in to situations where a plant can’t reproduce the results of the salesman’s demo.
That’s why I suggest you take five to read “Bringing Closure to Product Defects,” published on the Control Design website, February 13th, 2013. This article describes the process Phoenix Closures used to select a vision system that would be deployed across multiple lines. And believe me, it was a process and not a case of which salesman put on the best dog-and-pony show.
A pedant, like me, could argue there were weaknesses in the process. For instance, I think more time upfront defining needs would have been prudent, and some kind of scoring or ranking system could have been used to quantify relative performance. But perhaps all that was just cut from the article. I don’t know.
Bottom line: don’t buy the first system you see. Make the effort to figure out what suits you best. It’s always going to produce a more satisfactory result.

1 comment:

David Dechow said...

Hey - I keep enjoying the blog and try not to comment too often... Regarding, though, this end-user and the related article, I'm REALLY troubled by one of statements. One quote in the article says "The company had several brands of vision systems in use for a few years, and ... there were always problems. "We realized we had to figure out what brand and type of vision solution would work the best for us and why," The reality is that most often the problem is in the implementation, NOT in the product. It's HIGHLY unlikely that Phoenix Closure's previous inspection attempts were failures due to component selection. More probably it was due to failures in design, implementation, integration, customer buy-in, and vendor support. Now, that all may still translate into component selection, but I suspect some detail was missing from the article.