Camera Allows Automation to Do Many Jobs : Vision Control Cuts Costs in Processing
As the fast-moving conveyor belt carries the french fries to a packaging machine, an overhead camera feeds images of sliced and diced potatoes into a computer that, in turn, generates split-second determinations about the weight and length of individual fries.
An arcane use of an expensive technology?
Not really, explained Irwin Allen, vice president, finance, for International Robomation/Intelligence, the Carlsbad-based company that created the artificial vision control system that gauges the weight and size of the fries. “These plants process 1 million pounds of potatoes a day, and if you cut in half the number of bags being overfilled, there is the potential for enormous savings,” he said.
International Robomation’s vision system also identifies bags filled with the longer and heftier french fries that fetch a premium price from restaurant owners who eschew smaller fries.
The food processing industry application underscores how pervasive vision applications are likely to become as machines that can “see” replace employees who inspect products and monitor quality control.
Vision is “one of the elements of artificial intelligence being used as factories become more independent of direct human intervention,” Allen explained.
“These machines allow a reduction of inspection staff, which is a very significant cost benefit,” explained Hans Weber, vice president of San Diego-based IRT Corp., which produces vision systems that use radiation to penetrate the objects being inspected. “Human inspectors are also fallible, they’re not consistent. A machine concentrates all the time.”
Zero Defects
IRT’s automated inspection systems worked their way into general factory use after being introduced in defense and aeronautics-related industries, Weber said. “These zero-defect customers knew that if certain things failed they would kill people, or make expensive (space) missions fail,” Weber said.
One IRT system, developed under contract to the Army, scans the inside of loaded artillery shells for flaws in the explosive charges. Another provides a digitally enhanced picture that verifies the readiness of fuses used in military ordnance.
Industrial customers in search of quality control are also using vision devices in a wide range of applications. International Robomation’s devices, for example, guide robots that weld metal structures and scan optical lenses for imperfections that human eyes are likely to miss.
Another International Robomation system scans ZIP codes on the 350,000 to 500,000 packages and letters that Federal Express handles each night at its Memphis, Tenn., mail-handling hub, sorting “faster and more accurately than the (previous) manual system,” Allen said.
International Robomation, which began developing vision applications when the company was founded four years ago, initially concentrated its marketing on the robotics side of its business. However, when customers expressed the need for vision devices, the company switched its emphasis.
‘Enormous’ Potential
The company now expects vision sales to generate about 80% of its anticipated $12 million in 1985 revenues. “We want to double (sales) every 12 months or so for the foreseeable future,” Allen said. “The market potential is enormous.”
Much of International Robomation’s future growth will come from its role as a supplier for GMFanuc Robotics Corp., a joint venture between General Motors and Japan’s Fanuc Ltd., which expects revenues of $150 million in fiscal 1985.
International Robomation has an “exclusive agreement to provide vision for GMFanuc, and in about two years we expect it will be a major relationship,” Allen said.
Two years ago vision devices were on the market “but they operated at approximately one-tenth the speed of human beings and cost $80,000,” Allen said. “Today’s systems are 10 to 100 times faster than a human and they cost $20,000 to $40,000.”
Kenneth E. Years, president and chief executive of Monitor Labs, a San Diego-based company that generates the bulk of its revenues from pollution control monitoring devices, believes the low end of that market is going to witness “explosive” growth.
Monitor, which last year purchased Image Data Systems, an Ann Arbor, Mich., company that produces automated visual inspection devices, is concentrating on “low-cost, automated applications--the hundreds of simple jobs in manufacturing that involve making sure the nuts, bolts, holes and labels that are supposed to be on the product are indeed in the right place,” Years explained.
Years is trying to capture the market for automated visual inspection devices capable of “handling five to 15 parts per second, (devices) that go to work right out of the box and function in the factory environment.”
“There are different segments of the vision market,” Years said. “It’s like saying the word ‘computer.’ Many segments and applications pop up. There are many viable markets for visual inspection, and we tried to target one that had not yet been developed.”
IRT, which is at the other end of the cost spectrum, produces systems that use radiation to penetrate objects and determine if the structures are sound.
Historically, IRT has relied on its military manufacturing customer base for growth. Just two years ago, for example, only 20% of its vision-related systems were purchased by commercial customers. This year, almost half of the company’s business is generated by the commercial sector. “That’s a very impressive and rapid acceleration,” Weber said.
