Easier Machine Vision

If seeing is believing, then consider machine vision trends. The technology is getting easier to use and pushing toward higher resolution. These developments should lighten the burden on machine builders and users alike, and improve their bottom lines.

For instance, VMT, a division of Pepperl+Fuchs, offers PC-based systems. An advantage of this approach is greater storage, says Todd Belt, group systems sales manager for VMT. Lots of storage allows an image database to be constructed and used for such tasks as inspecting metal, which can be dull, shiny, oily or dirty on different days. A database can likewise prove useful when training a system to track a glue bead that varies in appearance from one batch of parts to the next.

"When it comes in lighter, we can add that image to our database," Belt says. "So when the glue changes back to the darker, it automatically knows to be looking for the darker. It learns this way, because of the database."

VMT also makes vision sensors, barcode readers and application-specific vision sensors. The company does not offer a smart camera.

Often, vision is retrofitted to a machine, notes Ben Dawson, director of strategic development at Waterloo, Ontario-based Teledyne Dalsa. The company kept that in mind for its BOA smart camera by making it small and capable.

Another aspect of the same guiding design philosophy can be seen in the software. The trend has been to make life easier for non-experts to implement a project by, among other things, creating easily grasped tools and terminology.

"If you want to measure this edge-to-edge distance, you use a caliper tool," Dawson says in illustrating this approach. "In the past, accomplishing the same task might instead have called for running a particular algorithm. Now we're speaking the language of the process engineer or the manufacturing engineer."

Ease of use is further enhanced by the company's smart camera, which supports many different communication protocols. Setup is via Ethernet, further simplifying its operation.

The processing speed of a smart camera is not as fast as that of a PC-based system, Dawson says. Thus, there are some applications for which the technology will not work.

Getting the optics and lighting correct is often the most challenging part of a machine vision project. Improving ease of use in these areas lies behind a new offering from Cognex. The company's In-Sight 7010 entry-level vision system features integrated lighting and autofocus.

Autofocus makes changeovers from one part size to another much easier. Previously, that required manual intervention. "Now with the click of an auto-focus button, one can adjust for the best focus value during part changeovers without needing to manually adjust the focus," says Narayan Subramaniam, principal product marketing manager.

It's even possible to preprogram several different focal settings into the device, allowing machine builders and their customers to set up recipes ahead of time, he adds. To extend its usefulness, the system was introduced with five different integrated lens options. These allow it to match requirements for working distance and field of view.

Cognex also invested significant engineering time to make its inspection tools easier to use, Subramaniam says. Its products also support various communication protocols.

At Matrox Imaging, the emphasis has been on improving the software that runs machine vision systems, such as the company's Iris GT smart camera. In particular, says Pierantonio Boriero, product line manager, this can be seen in the Matrox Imaging Library and Design Assistant. "The first is higher-powered and can run on either a PC or a smart camera," Boriero says. "The second requires less programming skill and runs only on a smart camera. Both have gotten easier to use and more flexible."

As for the future, easier-to-use and higher-resolution systems will help builders and users, Boriero adds. For one thing, it could be possible to cut costs, with perhaps 10 low-resolution VGA (0.3 megapixel) cameras replaced by three or four high-resolution 5 megapixel cameras.