A Weld Bead Profile Extraction Method Based on Scanning Monocular Stereo Vision for Multi-layer Multi-pass Welding on Mid-thick Plate

In the shipbuilding and marine industry, multi-layer multi-pass welding (MLMPW) on mid-thick plate plays an important role, but most of the actual production is still completed by manual welding. It is a big challenge to improve its intelligence level. In view of the current multi-layer multi-pass planning (MLMPP), a lot of simplifications are carried out. Limited to simplified weld bead shapes. MLMPP is difficult to implement more accurate intelligent welding technology. In the field of Intelligent Welding Manufacturing (IWM), Intelligent Robotic Welding Technology (IRWT) is the key of the IWM. Improving IRWT for MLMPW is of great value. The extraction of weld bead information is very important for more accurate MLMPP and its online correction. In this paper, scanning monocular stereo vision for MLMPW is used to reconstruct the weld bead. Through the slicing and filtering of the point cloud data, the profile of the weld bead surface is obtained, which provides a solid foundation for MLMPP and its online correction.