by Andy Wilson
Networked machine-vision systems allow both product inspection and the product-inspection process to be more closely monitored.
Recently, an avid reader called to discuss how machine-vision systems were being used to increase manufacturing productivity. During the conversation, the reader claimed that the machine-vision industry had created a mini industrial revolution in automated manufacturing. He also wondered whether the productivity gains attained by deploying machine-vision systems across the manufacturing industry had ever been measured.
Unfortunately, the answer was no. Despite machine-vision-system deployment at such well-known manufacturers as Purina Foods, Phillip Morris, KLA-Tencor, and others, measurements of the number of man-hours saved and increases in production obtained across the semiconductor, automobile, and medical industries have, to our knowledge, never been documented. But, as readers of Vision Systems Design know, systems integrators are developing automated manufacturing systems based on OEM machine-vision components such as cameras, frame grabbers, and software to inspect products and devices.
In these systems, networked systems are often used to send pass/fail decisions to robotic actuators so that reject mechanisms automatically remove defective products. Interestingly, however, such networked machine-vision systems are only used to provide information about the part being inspected and not the state of the inspection system. This, however, is not the only way to deploy machine-vision systems to increase productivity.
Recently, for example, while I was surfing the Internet, I uncovered a seemingly niche publication, Modern Machine Shop magazine (www.mmsonline.com). In it, Mark Albert, executive editor, describes in great detail how Cessna Aircraft Co. (Wichita, KS) has retrofitted two machine tools with a software-based CNC system to create a database of production events as they happen. Using the system, Curtis Cook, a Cessna engineer, remotely accessed the CNC to solve a problem without making a trip to the plant floor.
According to Albert, Cook called up the CNC from his laptop computer and called up its diagnostic page. An alert message indicated that an I/O point was not responding correctly. It appeared that a proximity switch was not closing properly, and that created an error in the control logic. Cook then sent a command that forced that switch to open and close several times. It seems that a metal chip was lodged under the switch. Forcing the switch to continually open and close, by remote command, dislodged the metal chip.
Although the system described did not use image processing or machine vision, it highlights the power of networked systems and shows how remote diagnostics can be used to increase productivity. In the development of machine-vision systems, much design attention is paid to the inspection of the part under test, such as whether it meets dimensions, colors, or labels. But, although such systems are networked, less design attention is paid to automating the process of machine diagnosis.
With the introduction of the Ethernet and low-cost 1394-based networked cameras, however, it will be a relatively easy task for systems developers to automate both the process of product inspection and machine diagnosis. While high-speed cameras may be used to inspect products on a production line, networked 1394-based cameras on the same computer network could be used to provide remote machine diagnosis. In this way, engineers, such as Curtis Cook at Cessna Aircraft, would be able to remotely monitor the results of product inspection and the inspection process itself.