As a custom machine builder, we have to include machine vision once in a while as part of the overall control scheme. We’re considering the pros and cons of just buying a turn-key, all-in-one vision sensor package vs. something more versatile and powerful, even if it means doing some of the component integration ourselves to have the in-house expertise. It’s the software that scares us about that. I’d like to hear some advice about both approaches.
—From September 2006 Control Design
Use Application-Specific Solutions
I’m not going to give advice on both approaches, as there is a new class of machine vision products that will meet your needs in machine vision for flexibility and little software work. These products are designed to be quickly set up for a specific set of machine vision tasks, such as metrology or defect detection. We call these products vision appliances to suggest their application focus and ease of use. These products are a good balance between a limited but easy-to-use sensor or turnkey vision system vs. a flexible vision system that might take months to program.
There are three design features that make vision appliances flexible and easy-to-use. First, each type of appliance is designed to do a limited set of machine vision applications. Second, because the application domain is known, we can build in high-level, application-specific knowledge, rather than you having to program this knowledge into the application. You can think of this as an expert system within the appliance that greatly reduces the time needed to set up an application. Third, the user interface to an appliance is highly graphical and uses terms and constructs familiar to the application. For example, an appliance designed to do metrology “understands” how to make measurements, specify tolerance, do calibration, etc., and the user interface exposes this knowledge as familiar tools such as calipers that you can drag-and-drop onto the image to make measurements. You can often be up-and-running with this kind of vision system in a matter of hours.
It is good that you are willing to do some component integration in-house to gain expertise and, I’ll guess, save some money. In addition to the vision computer and its software, you need to integrate proper lighting, lens, camera, trigger sensor, and positioning (sometimes called staging or fixturing) for the parts or item you are inspecting. Proper lighting amplifies the visual details you are interested in and suppresses uninteresting details. Selecting proper lighting takes experience and experimentation, but most lighting vendors will work with you to reduce the selection time. Vendor tools for selecting lenses and cameras can also reduce integration time. The trigger sensor tells the vision computer that a part or item is in the camera’s field of view, so take a picture and process it. Positioning ensures that the part or item gets into the camera’s field of view. Positioning often requires some expertise in motion control—and perhaps getting your hands dirty.
Ben Dawson, director of strategic development, ipd/Dalsa Digital Imaging, Billerica, Mass.
All-in-One Solution Often a Good Choice
The question is one of the production automation industry’s most frequent and difficult quandaries. All-in-one systems tend to be convenient, cost-effective, and simple-to-use, while PC-based custom systems provide optimal flexibility. Let’s look at the factors:
Going with an all-in-one solution means that integration time most likely will be reduced, and should troubleshooting issues arise, there only is one vendor to contact for support. These types of universal vision systems typically shorten the learning curve by providing an intuitive, on-board user interface that makes program configurations and troubleshooting easy for everyone. This not only will save you up front integration costs, but it also will allow your customers to take ownership of the system, which, in turn, will reduce your number of time-draining, post-installation service calls.
Regarding power and versatility, I believe you can find both in a turnkey system. Each year, more advances are made to prepackaged vision systems that allow users to successfully tackle today’s most demanding machine-vision applications. Couple that with the fact that most universal systems offer these features at half the cost of PC-based systems, and you’ll start to see the large cost advantages of turnkey solutions.
That being said, there are advantages to going with an open-ended, custom vision system. You typically can meet the specific needs of your customers, and on most occasions these systems can integrate directly into any existing software packages that you might be running on current equipment. This kind of flexibility can be valuable for some applications, but don’t let this make your decision. There are a large number of third-party suppliers that provide this same flexibility by acting as a bridge between turn-key systems and preexisting software packages.
In the end, if you’re looking to avoid programming and PC software, you’ll find that an all-in-one vision system will not only meet your power and versatility expectations, but will most likely exceed them.
Joshua Jelonek, project manager, vision and marking technology, Keyence Corp. of America, Woodcliff Lake, N.J.
Check out "The Best Of" The Answer to Your Problems, where we list all published questions and answers.