Assembly operations that require picking randomly placed components off shipping racks are tough to automate. But that's what two U.S. automakers wanted to do. Automation specialists from Comau Pico (Southfield, MI, www.comaupico.com) evaluated the project and recommended vision-guided robotic autoracking systems as the best solution. The ISRA systems they selected included cameras, cabling, LEDs, and MONO3D software running on a vision system controller.
The robots will remove sheet metal parts from racks for placement into fixtures for car-body assembly. With camera and lights mounted on the robot end effector, MONO3D will locate the part in six degrees of freedom using fiducials such as holes or edges, and guide the robot to that part in 3D space. The vision system extracts 3D location data by locating three or more features in an image, determining the spatial relationships among them, and comparing that information to stored 3D models based on CAD data.
ISRA's Kevin Taylor notes the camera's small footprint that allows it to ride on the end effector improves the overall system accuracy. Using a single camera both reduces complexity and saves money as well on equipment outlay, installation, calibration, and operation. Depending on application requirements, MONO3D can provide the flexibility to determine spatial resolution, speed, and throughput by accommodating a virtually unlimited number of additional features into its calculations or by limiting the 3D vision to 2D or 2.5D (X, Y rotation and a scaling of Z).