Vision-based deviation extraction for three-dimensional control in robotic welding with steel sheet

摘要

Effective multidirectional extraction of the deviations between the welding torch and the tracking point is prerequisite for three-dimensional control in robotic automatic welding. This paper presents a vision-based method to determine the logical tracking point and further to extract the deviations of the welding torch in two directions for robotic pulse metal active gas welding with steel sheet in real time. Images from passive vision sensors include two edges of the groove, the joint line, and complete arc. A scheme is proposed to accurately discern the joint line based on the local grayscale maximum within the region of interest using orientation feature maps. The geometric center of complete arc is used to reflect the position of the welding torch, and the intersection of the extracted joint line and the line that passes through the geometric center and is vertical to the former is regarded as the desired tracking point. It is transformed to the three-dimensional world coordinate via vision calibration techniques, and the deviations in two directions are yielded when contrasted with the real-time, recorded coordinate of the welding torch from the robot control system. The experiment results show the effectiveness of the proposed method.