Lowering costs, increasing mobility, and enhancing resolution -- sensing and machine vision today combine existing and new technology to meet the exacting demands of industry, an IEN survey reveals. "It''s really a matter of looking at trends in other businesses and seeing what can be applied in our business," states DVT Corp CTO Phillip Heil, Ph.D. "These days we''re using the Web for training, product introductions, and even online diagnostics." (For more, click here.)
Dan Holste, manager of product development/vision products at Banner Engineering, says that manufacturers are moving from "islands of automation" to a "systems approach, working toward interconnectivity throughout the entire plant. By linking these areas of automation, a manufacturer can work toward avoiding scrap product and positively influence process flow." In this method the "vision sensor can be applied as a process control element," Holste suggests. "Mating interconnectivity with vision systems will help to create a manufacturing environment that will truly reduce scrap product by properly adjusting product variables prior to a final pass/fail type of finished product inspection." (For more, click here.) (Pictured, Banner''s PresencePlus2 vision system.)
Manufacturers are shifting to distributed vision, which allows "feedback about why a part may have failed, which in turn can help the company achieve better process control, "according to Carl Gerst, product marketing manager at Cognex Corp. "This enables them to catch defects at the point of occurrence, before more value is added to a part." (For more, click here.)
Reed Switch Developments Corp marketing and sales manager Vicki Lanphier points to reed switches as a technology that transcended its early application. "Reed switches or magnetic proximity sensors were originally developed for telephone relay applications. Today, they are ideal for supplying high-speed digital input into microprocessors like counters, PLCs, and position sensors," Lanphier explains. (For more, click here.)
"Plant floor machine vision is being given a second chance," observes Ralph Tesson, director of worldwide sales at Coreco Imaging''s Intelligent Products Div. "Traditionally, sensors have offered many qualities that early machine vision systems could not, including reliability/robustness, ease of use, and cost-effectiveness. As a result, plant floor personnel have shunned machine vision systems because of the failure of these systems to provide the same qualities as other sensing devices on the production line." However, advances in hardware and software allow the machine vision sector to "design vision sensors that provide the same ease of use and reliability as other sensors, while providing greater functionality," Tesson believes. (For more, click here.)
Customers need convincing that "machine vision is a proven technology and that a $4,000 system today can do everything a $20,000 system could do five years ago," Heil adds. "DVT is addressing the issue with free training and product awareness through over 100 workshops a month worldwide."
Dave Edeal, displacement product marketing manager at MTS Sensors, notes: "The fundamentals of magnetostrictive sensing technology have not changed significantly in the past 20 years. However, there has been much refinement in the manufacturing and materials processes involved -- resulting in significant gains in sensor performance and reliability."
More Power, Smaller Packages
Sensor makers now pack more power into smaller packages. MTS Sensors'' ServoSensor, for example, squeezes feedback and a "microprocessor based closed-loop position controller in a standard sensor electronics housing," says Edeal, while the presence of on-board Fieldbus interfaces "in many microprocessors and DSPs, makes it easier to provide enhanced performance and functionality in our Profibus, DeviceNet, CANbus and Interbus sensors without the significant development efforts required in the past." (For more, click here.)
Meanwhile machine vision can now "be used as a functioning part of the manufacturing process," says Mark Sippel, vision product marketing manager at Omron Electronics. "For example, when a hole is being drilled, a vision sensor can gauge the depth of the hole as it is being drilled inspect the diameter for conformance and use the data to help predict tooling wear on the drill bit." (For more, click here.)
Adds Dwight Davis, a partner of Vision Controls: "Digital video capture and compression, coupled with Internet transmission and analysis, is expanding the power of currently used sensing devices. While these sensors can be highly accurate, their scope is limited. Use of video provides the whole picture, tying together the precision of available sensors with the overall view of the situation. Furthermore, software can enable hours, days, or even weeks of video observation to be logged, retrieved and analyzed with the depth of statistics and the context of the human eye. This technology does not supplant currently available techniques, but rather integrates sensors and process control PLCs with high-resolution color sensitive digital video." (For more, click here.)
"Software must become easier to use," believes National Instruments vision product manager Jason Mulliner. "As PC power increases, users need a configurable way to quickly create inspection solutions."
Reducing integration costs remains a big challenge. "For example, it''s one thing to generate pass/fail results with a vision sensor, but then how do you get the results into a control program for process feedback?" Mulliner asks. "The industry needs to make this type of integration even easier. If you want to get data into a spreadsheet, for instance, you should be able to leverage standards such as OPC instead of writing your own custom drivers." (For more, click here.)
Joshua P. Jelonek, technical support engineer at Keyence Corp of America, tells IEN that "sensor companies have to meet the challenge of creating devices that can handle high volume inspections -- I wouldn''t be surprised if systems that can inspect over 10,000 parts/min will be available in the near future." Many of today''s systems handle 5,000 ppm. (Photo shows Keyence''s CV-700 series.)
Users increasingly insist on easier programming and additional functionality. "Unfortunately, customers are becoming increasingly demanding of their low cost systems; they''ve got harder applications to solve and less money to spend to solve them," comments Jelonek. "Now our R & D department is faced with the challenge of sticking to our roots of easy programming and adding more inspections tools to solve more challenging applications." (For more, click here.)
Training stands out as another major issue. "As machine vision becomes more mainstream in industry," Sippel remarks, "education is critical to make sure applications are correctly solved the first time with proper information and understanding of the technology. Machine vision needs to become more of a known science, not black magic. This can be addressed by providing users with the information of how lighting is used and how the common vision tools can be applied to applications."
New Uses for Vision Systems
The use of vision systems continues to spread beyond traditional end-use markets, industry leaders report.
According to Jelonek, OEMs have "decided to ''pre-integrate'' the systems while the machine was being built. This way an end user can buy a complete solution to an automation problem." Heil says that the OEM market is "reacting to the end users'' request for 100% inspection and in some industries traceability. We''re seeing a lot more interest from large machine builders in several vertical markets."
Early on, demand for machine vision technology came from "industries that could justify the higher cost of implementing the technology," Sippel observes. "But with the development of lower cost vision sensors, under $10,000 -- some under $5,000 -- new end-user markets, such as transportation (inspecting shipping containers), food inspection (fruit and vegetable quality), and wood processing (plank cut conformance) are adapting the use of vision for verification and inspection."
Cost and ease-of-use concerns have lessened, states Holste: "Today, vision technology is moving into the OEM market due to its ease of use and decreasing price points.In the traditional end-user markets, vision is not only more affordable, but has reached an ease of use level where general plant personnel can maintain the vision sensor."
Coreco Imaging sees an "increase in semiconductor, automotive, and pharmaceutical requirements for inspection on the production line," according to Tesson. "This demand has also filtered down to the suppliers in these industries to implement quality control procedures and inspection systems in their plants as well." Meanwhile, "offshore manufacturing has become less risky due to the increasing use of machine vision and the quality control that it provides." These benefits are persuading "companies in non-traditional markets such as textiles" to install machine vision, too.
Greater processing power has led to a "steady increase in resolution and camera frame rates," Tesson continues. The next step? "The focus for manufacturers of plant floor machine vision solutions is shifting from providing a hardware-based machine vision platform with a box of software tools to providing dedicated, application-specific sensors designed to be cost-effective at performing specific machine vision tasks," he predicts. "Neural network technology will train the machine vision sensor to perform simple tasks that differentiate defective products from the production run and to interface directly to the reject mechanisms. The beauty of these neural network-based systems is that this training can be achieved very quickly," according to Tesson.
Advances in Software -- and Hardware
"You''ll see advances from DVT on the software side making the SmartImage sensors easier to set up and open to communicate with a wide variety of industrial components," adds Heil. Keyence''s Jelonek expects the future will bring a higher resolution at a more reasonable cost, and that "Mega-pixel resolution will most likely be the standard." Omron''s Edge Code technology "locates and isolates a feature or object by the edges to either search by them or isolate the edges and remove them to allow complex measurements with higher degrees of accuracy," Sipple notes, and "using and gathering data from entire vision networks over the office LAN, wired or wireless, or even the Internet, is coming."
Dave Edeal acknowledges that software drives MTS Sensors'' new products. "Our modular design separates the base sensing element and electronics from the output or ''personality'' module. Using this approach, we tailor the sensor output using highly flexible software programs. Here, we incorporate a significant number of critical interface parameters or create custom recipes for specific applications," notes Edeal.
Expanding PC power allows the industry to add more sophisticated imaging tools economically, Mulliner tells IEN. "Also, users demand easier to use software without sacrificing functionality. The ability to create a configurable environment that bridges to the fully featured capabilities of a traditional development environment is necessary to meet these demands," he says.
"By integrating data with high quality, high frames per second digital video, analysis data is easily correlated with the visual picture," according to Davis. "Using compression, streaming, and event cataloging, information for analysis by expert teams dispersed long distances across the enterprise enhances the process improvement technique," he explains.
Gerst sees advances in both software and hardware. "From a software standpoint, vision algorithms are constantly being improved," he explains. "For example, today you can get optical character recognition and verification software on low-cost vision sensors that rivals the robustness and overall performance of more expensive PC-based vision systems. From a hardware standpoint, today''s vision products have more rugged enclosures, and do a better job in terms of offering protection from harsh industrial environments, washdown, etc.," notes Gerst.
"As development progresses," adds Holste, "higher levels of integration, increasing speeds, and denser memories provide the platform to support the need for increased functionality in these products. The software of today is also much more transportable, meaning that it is reusable, making it much more cost effective and more flexible than ever before."
How does current sensing and vision technology fit into the lean paradigm? "Lean and flexible manufacturing is helping to drive machine vision technology," Sippel contends. "It is lowering personnel costs while improving quality and throughput. It is also allowing machine vision interaction to affect the manufacturing process without the use of humans, like in hazardous environments or locations where humans can''t be used."
And Mulliner notes that "manufacturers do not want to have to replace their testing systems every time a new product is released or a current product is slightly altered. The ability to have an open and flexible testing system that is scalable to new technologies is essential in order for manufacturers to continue cutting costs."
Tesson provides a last word of caution: "It is important for manufacturers of machine vision products to render them both simple and compatible with existing technologies. The new generation of intelligent sensors that require little programming is of little use if they cannot interface with the rest of the production line equipment without re-engineering. These devices can operate as standalones, and often can replace a PLC; however, compatibility with PLCs, Ethernet I/O, and other communication devices is also necessary to ease the integration."