Teach Less Robotic System for Deburring Workpieces of Various Shapes

Abstract

In the case of conventional industrial robot systems, operators have to write robot-language programs for each type of workpiece. This is an onerous task, especially when each workpiece has a different shape. In this paper, a teaching-less robot system for the finishing of two-dimensional workpieces of various shapes and thicknesses using computer vision is proposed. The robot system does not require shape information for the workpiece to be included in the CAD data or to be input by the operator. Each workpiece shape is acquired by segmenting edges into straight lines and circular arcs from the image data of the workpiece. The robot-language program for each workpiece is generated automatically from the workpiece shape data and finishing condition data. The effectiveness of the proposed method is verified by experiments using a newly developed robot system. This method provides a compact and inexpensive finishing robot system which reduces the programming timing.