Visual Torch Position Control Using Fuzzy-Servoing Controller for Arc Welding Process

Abstract

In this paper, we propose a fuzzy-servoing controller method for automatic welding. The proposed method uses a vision based arc tracking to find the initial points of the weld seam and to track them without a prior knowledge. Due to a serious melt down in the weld pool during the welding process, the method requires to control the welding torch in two directions, up-down and left-right directions. To perform these, an IR, two CCD cameras and two stepper motors by inference of fuzzy rules are used to control the movement of the welding torch tip. Therefore, the proposed method can accomplish different tasks such as welding a curved seam or moving into multi directions, while majority of autonomous arc-welding approaches are single-purpose so that they are designed to accomplish only one task, like welding in a direct line or a predefined arc seam. This method is applied on several workpieces and then the maximum error in both directions is shifted to zero in 30ms. This time is proper for workpieces with thickness in the range of [1-5] mm.