provided by: 
Faced with the imminent retirement of quality-assurance and inspection staff, many small- and medium-sized metal-manufacturing firms may start considering the technological equivalent of a late-inning "double switch."
National League managers try to gain an advantage late in the game by substituting a pitcher and placing him—usually a poor hitter—further down in the batting order.
On the shop floor, managers can do the same by bringing in a new machine-vision system solution to replace departing employees and reap the benefits.
That's possible now since the value proposition of the latest shop-floor vision solutions has improved dramatically. According to Ed Roney, Rochester Hills, Mich.-based development manager, Fanuc Robotics America Inc. (www.fanucrobotics.com), during the past 10 years the costs of such systems have come down 75 percent while their functionality and ease of use have risen significantly.
Newer solutions can measure or inspect things that humans cannot, such as examining the threads of precision screws used in medical devices or other highly specialized equipment.
Greater penetration of such tools among metal-manufacturing firms will enable them to catch product flaws and correct process errors earlier in the process. Such machine-vision tools are not just about quality assurance; they help boost cost containment and production efficiency through in-line, in-production error-proofing.
MARKET CONFUSION
First-time buyers may quickly become confused over the various names of the emerging vision solutions. A recent study from Ann Arbor, Mich.-based Automated Imaging Association (www.machinevisiononline.org) divided the market into three segments: smart or intelligent cameras, vision sensors and embedded vision processors, sometimes referred to as embedded vision computers or compact vision systems.
However, report author Paul Kellett concluded that there is "no universally accepted definition of this market. Different companies hold divergent views of the market and its segments."
In other words, the traditional labels of smart cameras, sensors and vision systems do not necessarily describe the scope and power of the underlying tool. As a result of recent technological advances and cost reductions, newer systems that on the surface resemble the equivalent of a Volkswagen beetle may in fact have Porsche components under the hood.
If labels are not the answer to helping companies choose a suitable vision system solution, perhaps system architecture is. "Smart cameras come in three major configurations," says Kellett. "But essentially, they are compact, integrated systems. If you opened up the housing you could not separate the different functions.
"ASMVs [application-specific machine vision systems] are more expensive, modular systems with a larger footprint capable of handling more demanding and more complex tasks. They are made up of separate parts linked together. They are a separate category. But both are considered machine-vision systems."
The technology involved in machine vision can get very complicated very quickly. But in practical terms, as the technology matured and the relevant chips, hardware and computing power became cheaper, the various solution types—sensors, smart cameras and embedded machine vision systems—have all started to meet in the middle. As a result, the single-purpose basic sensors that could only detect a pass/fall operation and the more sophisticated integrated top-end systems have crept closer together in price, functionality and ease of use.
Many of the new systems, whatever they are called, can provide pretty close to out-of-the-box functionality that will meet the quality-assurance needs of most firms. But these mid-range solutions are not a cure-all for identifying every shop-floor product flaw and highlighting every process error.
"Producers need to find a balance between what they need to inspect and cycle time," says Adil Shafi, president, Shafi Inc. (www.shafiinc.com), in Brighton, Mich. "If a process requires only two seconds of cycle time, that limits what can be inspected.
"Some defects are easier to detect than others. For example, labels are usually in the same spot all the time, and it is simple to confirm if they are there or not. But if the vision system has to try to find 'anything anywhere', it can become much more complicated and expensive."
Consequently, the challenge for smaller operations is to find an affordable way to benefit from the vast array of vision solutions coming on to the market. If the names of the solutions alone are no guide, companies must first analyze their major quality-assurance needs and match them against the available tools.
Metal-manufacturing firms' ultimate objective is to make quality products efficiently and cost effectively. It is not to have a world-class quality assurance and inspection system. Technology is always a tool, never an end in itself.
UPWARD TRENDS
From the AIA's Paul Kellett's point of view, "The more valuable the inspection process is to the company, the more sophisticated the software needs to be because it provides better information to the company to maintain the quality of its products and the efficiency of it production processes."
According to the early 2007 AIA survey, sales of the smart-camera segment are trending upward. The 2007 forecast called for a 10.5 percent increase. By 2010, it is expected to reach 12.9 percent. In terms of penetration by sector in 2006, general automotive was the leader at about 20 percent, while metalworking was far down the list at 4.7 percent. The AIA does not break down the responses by size of company.
However, it is more likely that smaller companies are not as active. One reason for that is the issue of scale. "Smaller companies tend to use smaller machines," says Endre Toth Hudson, N.H.-based director of business development Vision Components (www.vision-components.com). "Earlier, when such a vision system might cost $50,000, for a large manufacturer using a million-dollar press, adding on an inspection system was not much of a concern. But for a small firm using a $100,000 piece of equipment, it would not work."
Still, the major barrier was the high cost of such systems and, especially, the setup and integration costs. Toth estimates that they could be as much as five times the original hardware costs. In the past, large operators were able avoid most of the implementation costs because of their in-house engineering and IT resources. Smaller firms had to call outside consultants and implementation houses.
"You needed people with the proper expertise and the right responsibilities to do it properly," he says, "especially teaching other employees how to operate the system." As a result, for many years smaller players shied away from installing such systems because they were costly, complex and difficult to implement.
However, it is only recently that smaller firms have been able to choose from such a wide selection of suitable systems. "Many vision-system vendors are only just starting to discover second-tier markets," says Toth of Vision Components. "Before, they focused mainly on large users. To reach this new market, they are now offering their products through channel marketers."
These new vision systems enable smaller firms to update their quality-assurance practices to in-line, in-process inspection from the final inspection of finished products. Today, manufacturers must be able identify defective processes and products as they occur and fix them soon thereafter. Otherwise, their rework and scrap rates will severely damage operating margins and their reputations, especially if those products ever get out into the marketplace.
THE GOAL OF TRACEABILITY
Such vision solutions also help manufacturers monitor and control product- and process-related data that has become crucial now that customers are requiring product traceability. Producers must now be able to track shop-floor processing as well as product quality information.
John Keating, Natick Mass.-based, product marketing manager, Cognex Corp. (www.cognex.com), says today's vision technology can help track date/lot stamps to, again, ensure supply-chain traceability. "For example, if the printer jams when applying the information, the vision tool, spots it immediately, stops the production line so employees can fix the problem right away.
"Although there is nothing wrong with the product, some of the necessary production information is missing. If it is detected later, it makes it tougher for producers to identify how many items are missing the data. Then after they do that, they will still have to put the products through the line again. More important, if there is ever a product recall, the manufacturer can use that information to limit the size of the recall by isolating the questionable products more precisely."
Here is where the value of the vision system grabs the attention of senior executives. After the vision system identifies product defects and indicates the production-line failures that led to them, how can that information be used to improve quality permanently?
According to Toth, there are two possible answers. "One is to build a feedback loop into the manufacturing system so that when vision systems detect a flaw or defect, it corrects the problem automatically without human intervention. The second is to set up a formal reporting process to alert supervisors and other shop-floor workers about the problem so they can adjust the equipment or process immediately to prevent a recurrence."
MAKING VISION AND PRODUCTION INSEPARABLE
Toth continues, "The goal is to build machine-vision technology into the production system at the beginning rather than to include it as an afterthought at the end. Make the vision part inseparable from the production part. Use the machine-vision system as the 'heart' of the production system not simply as an add-on."
The data collected by the machine-vision system can help develop long-term strategic value. "By creating various quality-assurance reports indicating defects per shift, per employee etc.," says Keating of Cognex, "a firm can start to introduce Pareto analysis—20 percent of the causes create 80 percent of the problems—to products and processes so they can fine-tune and correct the most common or expensive glitches."
In the past many manufacturers, especially smaller ones, did not collect such data because it was too difficult or too expensive. But the new solutions make it all possible.
"Sometimes the ROI can be very fast," says John Keating. "One user got back its investment on the very first day, after the new system turned up a defect the company had never noticed before."
One of the emerging mid-market solutions is the Checker from Cognex. Aimed at new users, it just requires workers to plug it into a USB port and follow the Windows-based menu-driven instruction and run through the relevant list of functions. "Although it may look like a sensor," says Keating, "it offers a lot more because we have embedded more intelligence in the chips to make it easier for workers use. But we did not necessarily reduce the functionality available on our larger more expensive systems in the casing."
For example, Checker deploys a team of inspection sensors. Brightness sensors look for dark or light areas on the part. Contrast sensors look for areas on the part that contain both bright and dark areas that include date codes, threads and other features.
As well, pattern sensors understand what part features are supposed to look like, so that when a sensor identifies a feature that looks like what it was trained on, it proceeds with its job.
Such teamwork increases the Checker's flexibility. Unlike older sensors, it detects the presence of weld nuts in automotive components by tracking the frame over a wide area, and sensing the brightness of the nut. Before, photoelectric sensors could only do their job when the object was precisely positioned in the same place every time. That required greater support activities, hardware and programming to control.
In this way, newer solutions give manufacturers flexibility and efficiency—pillar's of today's lean, just-in-time world.
Editor's Note: Ken Mark is a freelance writer based in Toronto. Photo courtesy of Vision Components.
author: By Ken Mark