Sensing with Robotics Robots are nothing new in industrial assembly. For example, everyone has heard of the heavy yet highly precise one-armed soldering or painting robots of the auto industry. Such systems take over laborious and repetitive tasks, raising production quality and productivity. Still, most industrial robots have the disadvantage of being very capital-intensive and reprogrammable only with great effort. For that reason, they are only profitable when used for large serial-capacity production, and they are not particularly known for their flexibility. The trend of today’s fast-moving economy, however, lies more along the lines of shorter product life cycles, greater product variety and faster adaptation to new market requirements. More flexibility is the Holy Grail of industrial production. Pi4_robotics GmbH (Berlin, Germany), a robotics and imaging specialist, has developed within the framework of the EU-aided research project Assembly System Integrated Project (PiSA), a new type of robot that brings human flexibility to industrial automation. The so-called pi4_workerbot is a compact robot that can take on many simple, unrefined tasks which previously had to be done by humans. The pi4_workerbot has been conceived to learn new tasks or to change from one job to another with little retraining, much like a human colleague. Even though the pi4_workerbot is no humanoid robot, it does possess certain similarities to humans: its size—just less than 2 meters—and proportions approximate those of a human so that it needs about as much space as a person. Like a human, it has two highly movable arms, a head and eyes. It has no legs, precluding independent movement, but instead stands with full steering technology on a rolling platform, allowing it to be easily moved from one workstation to the next. The pi4_workerbot does require a power supply at its new workstation; otherwise, it brings everything else along. Thanks to cameras and power sensors being built into its arms, the pi4_workerbot can see and feel what it grasps and manipulates. Thus, it is equipped to take on sensitive joining tasks and to self-monitor the quality of its own work. SUPERHUMAN POWERS One feature of the pi4_workerbot is its extensive configuration of industrial imaging technology. Depending on the execution and formulation of tasks, it makes use of up to three digital cameras. For applications requiring spatial perception of the workspace, an optional time-of-flight camera is mounted in the middle of the head. For inspection tasks, two additional digital cameras from Allied Vision Technologies (AVT) with different illumination units can be attached to the head. The cameras in question are area scan cameras with FireWire IEEE 1394 interface, such as the little AVT Guppy or the high-performance AVT Stingray. The importance of easily integrated cameras cannot be understated. AVT’s FireWire digital cameras’ versatile selections of sensors and functions fulfill this need because they can be switched in and out of the robot system with ease. The FireWire interface’s integrated power supply is particularly convenient and easy to switch out, allowing for a operator-friendly plug-and-play functionality. The pi4_ workerbot is not fixed to a particular interface, however. For example, digital cameras with Gigabit Ethernet interface also can be integrated. ABLE TO LEARN Crucial to the pi4_workerbot’s flexibility is its intelligent software. Pi4_control is a software solution that combines imaging with machine control. Thus, inspection tasks can be programmed particularly easily and efficiently, such as having the robot grasp objects and move them within the camera’s field of view to achieve optimal test quality. Pi4_robotics and Allied Vision Technologies demonstrated the pi4_ workerbot application at the Vision 2010 show in Stuttgart, where the robot inspected chrome-plated Peugeot emblems for an automotive component supplier. The chromeplated surface generates mirroring effects and light reflections that pose a challenge for testing with machine imaging. Moreover, the complex, angular form of the lion on the emblem requires that the part be turned to capture every edge and surface. Exactly as a human would, the pi4_workerbot holds the product in front of its camera eye and moves it to search for surface defects from all sides. The pi4_control software coordinates not only the machine controls and the imaging, but excels as well with an particularly simple operator surface, thanks to which the pi4_workerbot can be reprogrammed to a new task quickly and without complication, even by trained personnel without extensive robot programming experience. The pi4_workerbot’s potential range of applications is vast. For example, it can execute assembly, testing or packaging tasks that could not have been automated before and were handled by unskilled workers. Thanks to the pi4_workerbot, such personnel-intensive tasks that for cost reasons would have previously been outsourced to low-wage countries can be kept on location in higher wage countries. Alternately, for workstations that place workers in peril—such as those involving the manipulation of chemicals or laser beams—the pi4_workerbot offers an interesting alternative. Vision Sensors Improve Wheel Fastener Productivity One of Maclean Vehicle Systems’ (MVS; Royal Oak,MI) many automotive fastener products is stainless steel capped wheel nuts. The nuts are assembled on an index machine that processes two parts simultaneously at each station. The fasteners must be inspected to Verify the nuts are properly threaded. MVS previously performed this fastener inspection with an eddy current sensor whose accuracy was less than desirable, at times resulting in expensive over-sort. “We looked at a number of different vision sensors,” says TJ Konieczke, manufacturing controls engineer for MVS. “Some were sensitive, intimidating and had a high learning curve like the vision systems that we have used in the past. We selected Cognex Checker vision sensors because they are very easy to program and operate. We simply put them in place, connect a PC to the USB port, train them on the part to be inspected and pick tools off a menu that inspect the critical features of the part.” Konieczke mounted two Cognex Checker 101 vision sensors looking down at a 45-degree angle about 5 inches from the two parts in the inspection station of the assembly machine. The parts are presented to the inspection station in a fixture and held in position by a springloaded keeper, so there is some variability in the height at which they are positioned. Konieczke set up the vision sensor by dragging inspection tools and dropping them onto the features that he wanted to inspect. First, he addressed the variable placement of the part by using an edge finder tool to locate the edge of the part. Then he picked a brightness tool to look for the light reflecting off the threads. He then set up the vision sensor to interface with the programmable logic controller (PLC) that operates the assembly machine. The PLC provides a static output that indicates when a fixture is in position with two new parts. When the vision sensor receives this signal, it captures an image and inspects the part. The vision sensor then sends an output to the PLC indicating that the part is either good or bad. Based on the inspection results, the PLC determines whether to send the part to the pass or fail bin. Most recently, MVS has standardized new applications on the Cognex Checker 232 vision sensor that provides high resolution and a wide field of view to inspect small features up to 1,600 parts per minute. The ability of the vision sensor to determine the position of the part and evaluate the brightness of the entire opening of the nut, as opposed to an individual point evaluated by a traditional sensor, has eliminated incorrect pass/fail decisions. The elimination of sorting costs quickly paid for the new vision sensor. The vision sensor also has increased confidence that not a single bad part will be shipped to the customer. “The latest generation of vision sensors is much more reliable and accurate while being just as easy to set up and operate and not much more expensive than traditional sensors,” Konieczke concludes.
Published by QualityMagazine. View All Articles.
This page can be found at http://digital.bnpmedia.com/article/Case+Study/593187/56785/article.html.