Robotic Pick and Assembly Using Deep Learning and Hybrid Vision/Force Control

Abstract

Pick up of featureless cylinders using deep learning and assembly using combination of visual and force feedback is discussed in this article. Object detection using YOLO technique is used to detect the object for robotic pick up. It was possible to identify the object even in cluttered environments. The assembly of the cylindrical object in a hole with tight clearance is also discussed. The method involves hybrid control i.e., position control along the assembly plane and force control along the direction of normal to the plane. The method is implemented using KUKA iiwa robot and the results are presented.