A new system solution developed by SICK in collaboration with a system integrator for robot work cells is enabling flexible use of robots in automotive body assembly.
Pick-and-place stations in automotive body assembly are still loaded manually due to difficulties involved in using a robot caused by fast cycle times, the complex geometry of sheet metal components and the high accuracy of placement.
Even when robots are used, the sheet metal components are either supplied in special racks or singulated mechanically, or the robots are equipped with elaborate 3D measurement technology.
Door skins, hoods and other large components are delivered in packs of 70 to the pick-and-place stations, where they are inserted separately into the robot work cell. Only the first sheet metal pieces are well positioned on the rack, since they tend to shift slightly in the course of picking action. While this can be corrected manually with a minor movement of the hand, the robot requires constant, accurately defined picking positions throughout.
This issue can be resolved by means of elaborate rack designs, which necessitates greater logistics effort and considerably higher costs. Alternatively, one can determine the position of the sheet metal pieces using complex image processing systems.
SICK’s new solution, PLM (Part Locator for Manufacturing) uses IVC-2D smart cameras along with two different lighting designs, LED area lighting and a laser cross projector.
The robot first travels to a defined measuring position and projects a pre-set laser cross on to the metal sheet. Based on its geometrical displacement in space, the IVC-2D smart camera calculates a correction frame, transmitting the data to the robot via Ethernet interface.
The robot subsequently aligns the gripper parallel to the metal sheet, correcting its distance and starting the pattern recognition for gripper positioning while simultaneously adjusting to the individual features of the sheet metal components. Following the picking of the sheet metal part, the process is repeated until the rack is emptied.
Laser triangulation is combined with pattern matching to detect geometrical features reliably independent of fluctuations in contrast and ambient light interference, allowing fast cycle times between 200 and 300 milliseconds per image frame and positioning at a level of accuracy less than 1mm.
The integrated intelligence of the IVC-2D smart camera allows teaching-in new objects as well.