Wednesday, May 27, 2009

The essential feasibility study

If you’re quoting a machine vision system – one that’s going to perform a specific task – how do you know what the job is going to take? If you’re sitting on the buyer’s side of the fence, how do you know the vendor can deliver on his promises?

There’s only one way for buyer and seller to reduce the risk, and that’s by carrying out a feasibility study. A good feasibility study will show how the proposed solution satisfies the requirements, so providing a sound basis for developing a quotation. First and foremost, this means engineering a lighting solution (which almost certainly allies with selecting a part handling and presentation method, if that’s not part of the project constraints.)

With lighting defined it’s possible to determine:
- The rest of the vision hardware – camera, lens and so on.
- What form the image processing/analysis algorithm should take (and therefore how much development time is needed.)

And as you can’t put together a robust quotation without knowing how you’ll do the job, it’s essential to grab real images of real parts and evaluate a number of different approaches.

Unfortunately this takes time to do, and time is money. Most integrators are small and can’t afford to invest the man hours that a thorough study demands. For this reason, too many proposals are written based on experience and judgment rather than on facts and data. The risks should be obvious: cover all the contingencies and quote too high, or go low and pray you can meet the requirements. It’s no better on the buyer’s side: are you sure the guy with the lowest bid can deliver a working solution?

You’ve probably figured out where I’m going with this. A feasibility study is key to reducing risk, and as a good one takes time the buyer should be prepared to pay for it.

Integrators: I hear you chuckling, but when did you last ask for money for such a study? Try it, you might be pleasantly surprised.

No comments: