Robots and vision system automate bearing assembly
NOVEMBER 19--FAG Automotive (Schweinfurt, Germany), a wholly owned subsidiary of FAG Kugelfischer Georg Schafer AG, has enlisted Grude Systeme GmbH (Sauerlach, Germany) to automate its assembly process using off-the-shelf cameras, frame grabbers, robots, and robot controllers.
NOVEMBER 19--FAG Automotive (Schweinfurt, Germany), a wholly owned subsidiary of FAG Kugelfischer Georg Schafer AG that manufactures wheel mounts and bearing assemblies, was using manual methods in its parts assembly, with operators unloading parts from boxes and placing them on conveyors for delivery to an automated assembly station. Karl Bywalez, a project manager with FAG, has enlisted Grude Systeme GmbH (Sauerlach, Germany) to automate the process using off-the-shelf cameras, frame grabbers, robots, and robot controllers.
Grude Systeme developed a PC-based system using two robots and a multicamera machine-vision system. A C75CE CCD camera from Sony (Park Ridge, NJ) captures images of a number of parts. The images are digitized using a Series 8100 frame grabber and PatMax software from Cognex Corp. (Natick, MA).
"Using Cognex's PatMax software, two bearing components are first localized and their real-world coordinate transforms calculated," says Stefan Angebauer, managing director of Grude Systeme. These coordinates are then transmitted over the RS-232 port of the PC to an Adept Technologies Inc. (Cincinnati, OH) robot controller that directs an RX130L robot from Staubli (Faverges, Switzerland). The robot then picks the two bearing components and places them on a conveyor belt.
Unlike the bearing components, the randomly oriented wheel mounts are located in trays within the workcell. To place these parts on the second conveyor belt, another Sony C75CE camera captures images of the wheel mounts that are also interfaced to the Cognex Series 8100 frame-grabber board. To speed the pick-and-place process, the Cognex PatMax software locates two different mounts in the left- and right-hand side of the tray. Both images are then compared and best-fit image coordinates are used by the Adept controller to command the Staubli RX130L robot to pick and place the part on the conveyor belt.
For more information, see Vision Systems Design, Dec. 2001.