Leslie Gordon is a former Senior Technical Writer on Control, Control Design and Industrial Networking magazines.
Machine vision systems have long played an important role in inspecting, identifying and guiding parts. Current systems provide an increasingly sophisticated integration of machine vision and robotics, boosting the number of automation options that can help find defects, sort products and complete other tasks faster and more efficiently.
Industrial cameras play a big part in this trend. For example, camera technology has continued to improve not only by providing higher resolutions but also by carrying IP67 protection ratings, says Jeremy Jones, marketing communications at Baumer. "Several years ago, 0.8 MP cameras were most commonly used, but they had low environmental protection ratings. Currently, 2 MP cameras are available that allow the development of more advanced applications with integrated robotics thanks to their IP67 protection and capability to withstand higher shock and vibration."
Applications that integrate robotics have demanding requirements for vision system robustness because the systems typically work in harsh production environments. Some of these applications even involve moving parts where the camera is mounted at the robot. The more stringent requirements for robustness have been addressed by advanced specific camera design features, says Volker Zipprich-Rasch, head of marketing and product management, Vision Competence Center, Baumer.
"For example, Gigabit Ethernet interface with power over Ethernet (PoE) allows single cable solutions, which are beneficial for applications where the camera is mounted at the robot and moved dynamically," says Zipprich-Rasch.
"Advanced robotic applications benefit from cable length up to 100 m, and the system robustness is increased overall. Also, lightweight yet robust cameras are available so the robot needs less force to move the camera. A sophisticated mechanical design might include an IP65/67-rated metal housing to protect components against dust. Because the positioning repeatability of the robot is important, the camera sensor must detect position accurately, as well as work under a wide temperature range. These features support image acquisition with long-term stable reproducibility for advanced robotic applications."
Improved bead tools have also increasingly entered the picture. "For example, the heavy use of adhesives and sealants in the automotive industry means the bead measuring tools available in vision systems support easy integration with robots that need to confirm presence, size and continuity of the dispensed material,” says Brent Evanger, senior application engineer, vision, at Banner Engineering.
Bead measuring tools detect and evaluate continuous flows of mechanically applied adhesives or sealants.
"In the automotive industry, a robot moves a part under an adhesive dispensing nozzle to lay down a track of glue in the specific pattern," explains Evanger. "When the operation is complete, the robot moves the part to a vision inspection station where a camera checks out the applied material, deciding whether the glue is in the correct location and if there are any areas where too much or too little glue was applied. Results from the bead tool include the average width of the material along the length of the bead and the number and length of skips or gaps where material was missing. The vision sensor can then make a judgment call regarding the quality of the applied bead, passing the part on to the next station in the assembly process or rejecting the part for re-work."
Although bead tools are not a brand new technology, before they were developed, the primary method for ensuring the correct application of dispensed material was to closely monitor the temperature and pressure of the material itself, hoping to detect the skips in application as irregularities in pressure, continues Evanger. "The problem was that sometimes the pressure and temperature can be correct, meaning the dispensed material is flowing freely, but it may be shooting out sideways, not even hitting the part. Mis-directed flows of this sort might not be detected by simply monitoring the pressure in the supply line, for instance."
Current vision systems also make wider use of industrial communication protocols. "Years ago, there were only a handful of serial protocols such as RS-242, RS-422, and RS-485," says Greg Raciti engineering manager, Faber Industrial Technologies.
"But Ethernet technology has changed the way vision systems communicate with robots. For instance, many of the popular industrial Ethernet protocols have been incorporated into modern machine vision platforms, including Ethernet/IP, Modbus TCP and Profinet, to name a few. Because a single device can communicate with many various nodes on the network, the old peer-to-peer connection is no longer a limitation."
Raciti adds that another advancement is easy-to-implement calibration routines. "Software tools allow the vision hardware’s coordinate system to be transformed into one that matches the robot system," he says. "Most modern robots are programmed to move in millimeters and not pixels. Also, the robot has its own x-y-z origin. The capability to calibrate vision into robot units means the robot controller is no longer responsible for converting units and offsets. Instead, the vision system can locate the centroid of a part to be picked and send the robot controller its precise location in the robot’s native units."
