Despite the plant floor tendency to use "sensor" and "vision" interchangeably, there has always been a real distinction between the two.
Cognex Corp director of product marketing George Blackwell notes that "machine vision sensors deal with images rather than discrete inputs such as force, strain, load, or temperature. As a result, vision sensors typically provide an order of magnitude more information than photo-eyes or ultrasonic transducers." (Pictured, Cognex''s In-Sight 4100 vision sensor, with dual processors, network communications, and remote head camera, withstands harsh environments.) (Click here for more.)
Ben Dawson, director of strategic planning at Coreco Imaging''s ipd Division, defines a sensor as "something that translates energy patterns into data. . . .A machine vision system includes one or more sensors, usually in cameras or part-in-place sensors." (Click here for more.) While these differences remain, advances in sensing technology may begin to blur the distinction.
Current Challenges
The machine vision sector has been zooming in on ease of use and miniaturization challenges. "In the past, users were concerned with the setup of a single vision sensor," observes Blackwell. "Now users need to configure and manage an entire network of vision sensors." He tells IEN that "we will continue to see a migration of high-end tools such as edge detection into the lower end." Reducing size and components "improves reliability and reduces costs," Blackwell adds. "Consequently, vision sensor suppliers continue to decrease the size and weight of cameras for use on moving equipment, or for embedding more deeply into machines."
Increased focus on inventory control makes greater versatility an imperative. "With inventory levels getting increasingly leaner, the sensors that are specified and stocked must be able to perform in a variety of demanding applications," states Jeff Allison, product manager/photoelectric sensing at Pepperl+Fuchs Inc. (Click here for more.)(Pepperl+Fuchs''s VCS110 Color Sensor, pictured, allows programming of up to 10 colors with one sensor, and can be programmed to either detect or ignore minute variations in color.)
Another issue facing the sector is "the tendency by some machine vision manufacturers or their distributors to ''oversell'' their vision sensors," according to Chuck Bourn, director of marketing communications, PPT Vision Inc. "We have heard from a number of end users who have told us that their smart camera is now being used as a ''doorstop'' because it couldn''t perform as promised." (Click here for more.)
Meanwhile, recent developments at the IEEE 1451 Smart Sensors working group are "leading to some innovative solutions" in the accuracy, cost, and configuration area, says Ryan Wynn, DAQ systems product manager at National Instruments. "With the soon-to-be ratified 1451.4 standard, sensors now have the ability to store specific information (i.e. model number, manufacturer ID, linearization constants, etc.) that will both ease the configuration process and improve measurement accuracy," Wynn explains. "More specifically, the 1451.4 standard specifies a standard TEDS (Transducer Electronic Data Sheet) template for each type of sensor, which is stored on an EEPROM embedded in the sensor itself." With this information, any TEDS-enabled data DAQ will recognize a sensor "when it is connected." (Click here for more.)
Envisioning the Future
Gus Vargas, machine vision product manager at Aromat Corp, believes that "future systems will be required to detect physical attributes in a similar way as human operators do, focusing on features such as color, size, shape, and distinguishing markings as part of a single and specific product, rather than performing multiple and independent inspections." By using AI software within compact CPU platforms and mega-pixel resolution cameras, systems will "distinguish complex patterns and features for multiple products, while reducing the hardware cost and remaining affordable." (Click here for more.)
Advances in software have put vision vendors in a difficult position. "There''s a trend toward more software capability at a lower cost," notes Joshua P. Jelonek, technical support engineer at Keyence Corp of America. "Now many customers are demanding high end performance at low end prices." (Click here for more.)
The experts disagree about the future of wireless in vision. "The role of wireless may grow for machine vision as applications in the non-traditional markets grow," says Mark Sippel, vision product marketing manager at Omron Electronics. "For example, container inspection in a shipping yard or freight depot may be better served by using wireless to transmit data like container information than trying to run wire to the collection system." (Click here for more.)
But Jason Mulliner, vision product manager at National Instruments, believes that the "bandwidth just isn''t available for streaming large images." (Click here for more.) And ipd''s Dawson contends that plant floor electrical noise "can corrupt the signals." Jelonek cites high costs as a roadblock, while Blackwell believes wireless will be limited to "operator interface applications such as portable display technologies." Still, potential advantages, including increased flexibility, could drive manufacturers toward wireless networks at some point.
Shrinking Size, Increasing Versatility
Allison sees the industry shrinking photoelectric sensors while increasing their range and adding versatility. "Today, a sensor with a housing of less than 1/2 cubic inch (8 cubic cm) can operate at a range of over 30 feet (9 m)." Photoelectric sensors now include "multiple mounting possibilities," Allison notes. He also points to advances in color and contrast sensors, including "fast and simple setup, most often through push-button or a remote teach," and greater intelligence.
Smaller, more power efficient temperature and humidity sensors in battery-powered dataloggers, as well as system-based loggers are coming into the market, says Joanna Phillips, product marketing manager at Onset Computer Corp. "While standalone dataloggers are suitable for a broad range of applications, there may be situations where you need more of a centralized data gathering mechanism for measuring a number of parameters," she notes. "These datalogger systems are also battery powered, and allow you to configure the system with any combination of available sensors." (Click here for more.) (Affordable HOBO LCD datalogger from Onset Computer, pictured, displays temperature / humidity and includes programmable high/low limit alarm setpoints.)
Machine vision technology has spread from the semiconductor and electronics industries to automotive, packaging, and other sectors as it matured. "The key drivers include greater line speeds, production flexibility, and product traceability," observes Blackwell. "Human inspection and intervention can''t keep up across all industries where line speeds are on the rise. The need for production flexibility for more frequent changeovers has increased in many industries because the trend is to make a number of different products on a single production line. Greater product variability increases the need to error-proof the production process."
Nanotechnology eventually may play a role in sensing, according to Daniela L. Carrillo, Sensors Group research analyst at Frost & Sullivan. Nanosensors can detect and respond to "physical stimuli with dimensions on the order of one billionth of a meter." Current R & D focuses largely on the biomedical area. Future users include the semiconductor, process control, and pharmaceutical sectors. "One of the biggest challenges for nanosensors is the ability to interface between nanoscale devices, microsystems, and macrosystems," Carrillo tells IEN. "Nanosensors will need to convert optical, chemical, biological, and electrical data into signals that can be transmitted within nanosensor systems, and that can be acquired by data acquisition systems and computers that allow for human interaction and analysis." (Click here for more.)
Can sensing and machine vision contribute to a lean solution? No doubt about it, according to Jelonek. "I don''t think it''s a matter of can sensing and machine vision be a part of a lean/flexible manufacturing solution, they are a necessary component in improving quality control standards and increasing operational efficiency. Oftentimes the initial outlay of cost to install a vision system is covered within the first few months of production, and manufacturers are starting to notice the bulge in their bottom line profits."