Abstract :
Various types of robots are utilized in industrial production systems in order to increase the productivity, reduce labor costs, and keep regular production. At present, however, human beings can perform a variety of tasks than the robots. To improve the dexterity, we treat robotic assembly of modular fixture parts used in the metal mold productions. To assemble the fixtures, we use a low cost 4-finger 12 d.o.f. hand. A camera is mounted on the palm, a 6-axis force sensor is mounted on the robot wrist, and OpenCV is used for image processing. In this paper, we investigate a grasping and insertion task of a bolt to a hole. And we investigate a grasping and insertion task of a jig part to another jig part. In these tasks, uncertainty of grasping by the hand occurs. To overcome the problem, we utilize the force sensor and the camera image. Handling and insertion algorithms of the bolt and the jig part are proposed and experimentally verified.
