Development of new teaching method for industrial robot using visual information

Abstract

Today, automation of various work using robots is progressing in the industrial field . For robotized automation, we must teach motion to industrial robots. However, small and medium-sized enterprises which make various products in relatively small scale cannot introduce such robots because teaching takes long time. In this research, in order to solve this problem, we have proposed and developed a new motion teaching method which can be done easily and fast. Motion teaching takes long time because of robot installation error and absolute position error. Therefore, robot motion teacher needs long time to do remote teaching. Thus, we propose a new motion teaching method that can ignore influence of these errors. By marking work places and capturing work targets using RGB-D camera attached to robot's end-effector, we generate motion paths in the robot coordinate system and robot can ignore influence of those errors. In order to verify the feasibility of the proposed method, we constructed a recognition algorithm for work subjects, work places and working environments using RGB-D camera. After that, the time required for motion teaching was compared between the proposed method and remote teaching through actual machine experiments. From the results, we suggest that proposed motion teaching method can perform in about half the time of remote teaching.