Abstract
We propose a method to generate robot-motions for product assembly in 3-D virtual workspace. In our method, an assembly operation is built as a combination of elemental tasks including robot-motion sequences. These elemental tasks are abstracted to be applicable for any products, and their motion information does not include any concrete positions in 3-D space. Then, teaching points to perform each of elemental tasks are numerically calculated by the motion-sequence, and 3-D models of products and facilities. These teaching points are merged and ordered in a sequence of elemental tasks. Finally, robot-motions for an assembly operation are generated from these points by inverse kinematics. An experimental result of assembly operations for a commercial laptop computer shows that our system reduced the teaching-time to 25 percent compared to manual operation using a teaching pendant.
