By Alice Shepherd
By Alice Shepherd
In the past, robotic vision was confined to the laboratory and perhaps a few custom applications. Nowadays, robust robots are using sight in non-fixtured, unstructured factory environments, thanks to the integration of artificial intelligence with traditional industrial robotics. A camera, robot and robot controller are linked to a computer, allowing the robot to see, move and react much like a human being. Many automotive parts manufacturers have embraced vision-guided robotics (VGR) and integrated the technology into their factories.
“Manufacturers have been spending billions of dollars each year on custom-made, complex, accurate fixturing devices to present parts to blind robots for processing,” says Owen Jones, chief executive officer, Braintech, a British Columbia-based software company that develops vision-guided robotic systems. “Unless parts were held in exact, perfectly repeatable positions, robots were unable to process them because they could not see them. Other jobs, such as loading and unloading, could not be performed by robots at all because they could not locate their position or orientation in space.”
Robots with seeing power have come a long way since then. “Vision-guided robots combine the best of both worlds-man and machine,” says Jones. “They possess a human’s ability to see, adjust to changing environments and plan ahead. When combined with a machine’s accuracy and untiring efficiency, significant manufacturing efficiencies are realized.”
According to Jones, some manufacturers will realize immediate capital cost reductions and return on investment (ROI) with VGR systems. Many VGR systems available on the market today are affordable and can replace costly custom-made fixturing equipment with equally high maintenance costs. Also, since robots are uncompromising playback machines that do not make mistakes, forget assignments, or contaminate parts, they provide accuracy, precision and high quality output. Other potential advantages of VGR include reduced labour costs, improved efficiency and productivity, and enhanced quality of life for workers since robots can perform the dull, dirty and dangerous jobs-or the “Three Ds”-that workers don’t like to perform.
At a major transmission manufacturer in the Northeast, for example, 30 heavy transmission housings arrive at the robot workstation, arranged in two layers in an eight-inch by five-inch steel basket. A plastic tray separates the layers, and every housing is at a different angle. The components in each basket can be two different designs, suitable for four-wheel or two-wheel drive transmissions. Before the robot can grab one, remove it from the basket, align it with a custom rack, and place it on the rack for transport to a machining station, it must visually identify which components are in a given basket. Jones explains, “Our system captures an image of the component via a robot-mounted, single-lens camera, transmits the image to a PC, converts pixels to millimetres, matches the object to one of those in memory, and then informs the automated controllers for the robot and the entire work cell so work can begin.”
At a powertrain plant in New York, when paired engine heads emerge from casting (two attached) heads, they are loaded onto a monorail system where they are arranged in an imprecise fashion-10 degrees in any direction and several inches out of alignment. Traditional blind robots would crash into the heads, generating costly downtime and expensive scrap. Now, a robot with sight is sent to a position about three feet from the engine part. The software directs the robot arm, which is equipped with a camera and light, to capture an image of the part and send it to a PC. “Trained” to recognize the part, the software instantly calculates where it is located in three-dimensional space, relative to where it would be if it were perfectly aligned. The software orders the robot to move its arm to new coordinates so that it is aligned with the part, and work can begin. In a matter of seconds, the system moves, tilts and finely adjusts the position of the robot. “A casting plant is a particularly challenging environment for a vision system [because of the] dust, gases and heat from molten aluminum. [VGR] works perfectly even under these difficult conditions,” says Jones.
Alice Shepherd is a Southern California-based trade press writer specializing in technology and management topics. To contact her, call Trade Press Services at 805-496-8850.