Recognition and identification the position and location of tooth saw butt joint shape

Abstract

This research presents a new approach vision-based to recognize and locate the tooth saw joint position in butt welding joint. The research came out because some joints are quite difficult to recognize the butt welding position and location especially in a corner side or which one belongs to the tooth saw shapes. Thus, the main objective of this paper is to synthesize a novel approach that is related to weld joint identification and detection, and then develop a new approach based on contour regions for recognition and identification of butt welding joint in robot welding. The methodology involved in identification and detection approach was implemented in two stages: (1) pre-processing and (2) segmentation. The butt welding joint for recognizing position from the starting, auxiliary 1, auxiliary 2, and ending of the joint shape was obtained from contour and region points. The recognized process was done in two stages: (1) a new approach based on the shape of weld joints and (2) weld joint representation in three approaches which are (1) middle point of each end of two lines, (2) maximum or minimum row of the lines depending on auxiliary points, and (3) intersection point between the two lines. The findings of the approach show that the largest image-matching error happened in approach (3); however, this approach has a smaller matching error compared to the actual position. The expected output is that a weld joint can allocate the position from the start, auxiliary 1 and auxiliary 2, and end points in x-y coordinates to instruct the welding robot to weld. This research will significantly increase the welding process of mass production from low to medium volume manufacturing either repair or maintenance work should be faster than welding parts manually.