Robotic Manipulation of Highly Irregular Shaped Objects: Application to a Robot Crucible Packing System for Semiconductor Manufacture

The basic technology for a robotic system is developed to automate the packing of polycrystalline silicon nuggets into a fragile fused silica crucible in Czochralski (melt pulling) semiconductor wafer production. The highly irregular shapes of the nuggets and the packing constraints make this a difficult and challenging task. To address this task, key research areas are identified, developed, and integrated. In this system, nuggets are grasped by a three-cup suction gripper and manipulated with a seven-degree-of-freedom SCARA manipulator. An optical 3-D vision system, based on active laser triangulation, measures nugget and crucible profiles. A model-free Virtual Trial and Error packing algorithm determines optimal nugget placement in real time. A hybrid position-force control scheme has been implemented and tested for physical nugget placement. The simulation and laboratory tests show that the system has the capabilities of meeting high production rates, achieving high process constraints, and maintaining cost effectiveness that exceed levels obtained with manual packing. The results suggest that model-less robotic sensor control systems can be effective in manufacturing applications. The key contribution of this paper is to show that robot systems can be effectively used to manipulate highly irregular shaped objects in the context of real commercial manufacturing processes.