Identifying Welding Skills for Robot Assistance

This paper aims at identifying the differences between skilled and unskilled welders by analyzing the position data obtained with a 3D motion capture system. Using the captured position data of various markers on the arm of the welder and the welding torch, the tip point position of the torch, the elbow angle and roll angle of the human arm are constructed. These constructed values are analyzed. The time domain analysis reveals that a skilled welder has better control over 1) the movement of the torch tip in the perpendicular direction to the welding line, 2) the height of the torch tip from the welded material, and 3) the speed of welding through the welding line. The frequency analysis shows that the amount and amplitude of vibrations are larger with the unskilled welders. The statistical analysis reveals that the difference of variation of the parameters between the skilled and unskilled welders is significant.