Multi-Layer Welding Path Correction of Medium-Thick Plate Based on Vision System

The multi-layer and multi-pass welding (MLMPW) process plays an important roll in the field of medium-thick plate automatic welding. To solve the problem of the robot self-adapting to the deviation of the weld position and the groove size due to the influence of the workpiece positioning and the welding deformation in MLMPW, a novel method based on the laser vision system to correct the path deviation is presented in this paper. Firstly, an image processing algorithm with image defogging technology is developed to deal with the interference problem of laser reflection on the metal surface. Then, the MLMPW path planning is performed based on the feature points extracted from the image. Lastly, a multi-layer welding path correction (MLWC) method is introduced to enable the robot to automatically scan the actual weld profile information of the previous layer and correct the welding path position of the next layer. The experiments demonstrate that the proposed MLWC method can effectively reduce scan times and eliminate the influence of welding deformation, and improve the adaptability of industrial robots to MLMPW.