Weld seam profile detection and feature point extraction for multi-pass route planning based on visual attention model

Automatic multi-pass route planning is one of key technologies for thick plate in robotic metal active gas (MAG) arc welding. In this research, a scheme for the extracting feature points of the weld seam profile to implement automatic multi-pass route planning, and guidance of the initial welding position in each layer during MAG arc welding, is presented. It consists of two steps: first a vision sensor based on structured light is employed to capture laser stripes and molten pools simultaneously within the same frame, and the laser stripe, forming the weld seam profile is detected by a visual attention model based on saliency. Then a methodology of polynomial fitting plus derivatives for feature point extraction of the weld seam profile is suggested. With respect to the effectiveness of highlighting the laser stripe, the proposed model is much better than the classic ones in this field, whereas the feature point extraction methodology in this paper outperforms typical template matching. Finally, the performance of the proposed scheme is demonstrated on different weld seam images captured in different layers and different welding experiments. New vision sensor captures laser stripes and molten pools simultaneously.A visual attention model for detecting weld seam profiles is proposed.Three kinds of feature points of weld seam profiles can be extracted.Different images captured in different layers validate the proposed methods.The extracted feature points can be used for multi-pass route planning.