Machine vision is becoming a key component in flexible manufacturing, as accuracy and repeatability reach new levels and costs decline. "Because vision systems can decrease manufacturing costs by identifying faults before value-added processes occur through in-line inspection, more companies are employing them on the factory floor," notes Jason Mulliner, National Instruments machine vision product manager. (Click here to read complete interview.)
"There's an increase in the use of vision for more basic inspection requirements because of vision's price and ease of use," agrees Mark Sippel, vision product marketing manager at Omron Electronics.
The OEM Market - and Beyond
As a result, machine vision continues to spread beyond the OEM market. Chuck Gillingham, senior manager, industry marketing at Cognex Corp, tells IEN: "The packaging of vision systems has changed in terms of traditionally being geared toward OEM type applications where vision may have been a PC board plugging into a larger set of OEM machinery, to where it can be offered as a more self contained, integral package and more familiar in appearance to and handling by end users." In the past, OEMs "typically looked at vision as a subsystem to their overall machine design," Gillingham elaborates. "Today, we're still taking advantage of conventional vision processing; only we are putting it into a package most end users feel comfortable with." The company's In-Sight vision sensors use a vision spreadsheet interface, for example. (Click here to read complete interview.)
"Machine vision systems started out in the hands of integrators and OEMs," according to Michael E. Williams, market communications manager at DVT Corp. "The systems of 20 years ago were difficult to set up and finicky to keep going." With the introduction of the smart camera and Windows-based FrameWork user interface, "the market has shifted to traditional end users." Technical innovations will "impact the price, performance, and range of applications that can be reached by machine vision technology," says Williams. "Over the past five years, prices for complete vision systems have dropped from $10k to under $3k. At the same time, high resolution (Mega Pixel CCDs) and color systems have entered the low-price market." DVT's high-resolution SmartImage Sensors, for example (as shown in the picture), detect color discrepancies at fast speeds and low cost.
Globalization will also drive this market. "As more and more manufacturing is leaving the U.S., machine vision will be counted on to make sure quality is still present in cheap labor markets," Williams predicts. (Click here to read complete interview.)
Nello Zuech, president of Vision Systems International, also cites quality control. Zuech tells machinevisiononline.org (the website of the Automated Imaging Association): "In the electronic industry there are compelling reasons to adopt X-Ray-based machine vision systems... Detecting rejects immediately after a value-adding step such as soldering can significantly reduce warranty costs, field repairs, etc." Continuing miniaturization and "finer line widths of printed circuit boards [make] solder joints ever more critical as they in turn are smaller and closer together, making them more difficult to inspect by people." The subjectivity of human inspectors also leads to "a high incidence of false rejects[and] excessive rework and inevitably that rework results in damage to the board creating scrap," Zuech observes. (For more, click here.)
Impact of New Technologies
New technologies continue to expand machine vision flexibility and efficiency. "Camera Link is a new digital imaging standard that provides fast data transfer along with standard and simple cabling," says Mulliner. "This gives vision integrators the ability to avoid custom cables and configurationsConfigurable software that leverages the ability to implement customer-defined algorithms is essential."
Meanwhile, software algorithms "are being created constantly," Mulliner observes, and "the user experience is being enhanced by the introduction of configurable software. No longer must the machine vision user program in complex environments. Simple machine vision tasks can now be solved with simple menu-driven programs. Complex integrated solutions require advanced hardware features to ensure synchronization among different hardware devices. Technologies such as RTSI which allow timing signals to be passed over dedicated buses to other automation devices decrease latency and improve throughput."
Cognex's Gillingham points to interconnectivity as another area of innovation. "What we've been able to do is not only offer the physical communication ports necessary to link vision with other devices but also to adhere to existing open standards in the factory automation marketplace such as MODbus TCP, Ethernet/IP, ProfiBus and others We're also developing self-guided wizards for our vision sensors for guiding users in setting up communications," Gillingham states.
"Machine vision is becoming more simplistic and user friendly," adds Sippel. "You'll also see innovations in advanced detection algorithms. For example, Omron's Edge Code technology will allow for complex detection and measurements using a method of advanced edge detection for model inspection and defect inspection."
Sippel also points to "improvements in the hardware of vision systems, like increased processor speeds and the availability and lower cost of high-density memory." These enhancements allow "software more freedom to add complexity and improved interfacing" to vision systems. Sippel contends that the "greatest improvement thus far with machine vision systems is with throughput. The speed of a system doing complex measurements has increased dramatically."
DVT's next generation product, set for release in late 2001, will use SVGA CCD chips for high resolution imaging and color CCDs, allowing color monitoring and color object finding, software support for IP over Ethernet and TCP/IP communications. Meanwhile, the Hitachi SH4 processor in DVT's Model 540 camera now "boosts the digitizing time to 13 milliseconds, which allows the vision sensor to be used in even the fastest manufacturing lines," notes Williams.
Concerns and Challenges
Quality and accountability concerns continue to drive the electronic integration of the plant floor. "DVT's Visual Data Network (VDN), a network of Ethernet-linked SmartImage Sensors, reports such statistics as units produced, orders shipped and defects found directly into the plant's ERP systems via the Ethernet," according to Williams. "Eventually, plants will make this factory-floor, visual data available to their suppliers over the Web."
Gillingham believes that "lean manufacturing is a key driver" in machine vision as part of a move toward mass customization. "Specs may change from part run to part run, so there's more potential time lost due to set up changes. Through networking multiple vision sensors across the packaging line, you now have the ability to change inspection parameters from a central location," Gillingham comments.
"Machine vision is excellent for lean manufacturing," Sippel agrees. "It plays to the role of automated quality control, as opposed to manual quality control. You have lower cost, improved repeatability and better quality."
The availability of off-the-shelf components also belies the lingering perception of machine vision as a high-cost solution. Jeremy Govier, senior applications engineer at Edmund Industrial Optics, explains that "off-the-shelf optics provide a fast and inexpensive way of prototyping a system, and may be more economical for the production run of the system as well. Because lenses are made in batches, a single custom lens costs much more than an off-the-shelf lens and takes much longer -- 8 to 10 weeks -- to arrive. Off-the-shelf optics, on the other hand, are made in quantity, in continuing production, and are designed into standard matrices." (For more, click here.)
Machine vision faces several challenges. The rapid pace of innovation makes standardization an issue, for example. It is unclear which digital camera standard (Camera Link, IEEE 1394, or possibly USB 2.0) "will become the de facto standard," Mulliner notes.
Sippel believes that the "major concern facing machine vision is primarily connectivity -- information transfer, networking, and open compatibility between systems or equipment."
And the increased dependence on this technology for global quality control poses installation and maintenance problems. "Machine vision will have to be easy to set up, locally supported, affordable and have the capability to have troubleshooting done from a remote location," Williams comments. DVT offers a global network of automation solution providers to service local accounts, free training, and "Ethernet connectivity built right into every SmartImage sensor," adds Williams. The latter allows users "to perform online diagnostics from any Ethernet connection in the plant or from any Internet browser around the globe."
The expanding customer base for vision mandates an increased focus on education. "The successful companies will be the ones that train their end users to maintain systems as well as discover new applications in their plants," Williams tells IEN. DVT's free VirtualTOUR CD "contains over 40 hours of training on vision sensors, lighting techniques and lensing instructions," Williams says.
