Research On Recognition Method Of Right-angle Fillet Weld Based On Line Structured Light

In recent years,with the transformation and upgrading of key industries in the national economy and the cultivation and expansion of strategic emerging industries,the scale of China’s intelligent manufacturing equipment industry has grown rapidly,with welding industrial robots being an important part.The welding robot is widely applied in the industrial manufacture of automobile,aeronautics and astronautics,ships,and large equipment.The welding joint recognization and guiding for the welding robot is the main technique for realizing the intelligent welding.With the rapid development of the machine vision technology,the line structure light 3D(LSL3D)vision sensors are widely used to guide the welding robot to realize automatic seam tracking,which has the advantages of low cost,high efficiency and high precision.The right-angle fillet weld is one of the main fillet welds.The right-angle fillet welds have the characteristics of special shape,smooth surface,and interference from welding noise.Therefore,the identification of right angle fillet welds is a technical difficulty in weld seam recognition.The current recognition method has the defects of low timeliness and insufficient applicability,and week antijamming capability,which needs to study deeply,and developed the novel LSL3 D sensor.The main contents include:(1)The preprocess of the image of the right-angle fillet weld and center of the laser image extraction algorithm.Using the image preprocess method to flitering the image of the right-angle fillet weld.The extraction algorithm of the laser center is applied.An right-angle fillet weld angle point recognition method based on the second order differential of the image outline is proposed.The second order difference value is calculated for the extracted laser center,and the weld joint characteristic point are selected by the non-max value restraining method.(2)Calibration of the LSL3 D vision sensor.The calibration of LSL3 D vision sensor includes the camera calibration and the light plane calibration.Camera calibration obtains the internal and external parameters of the camera,so as to realize the conversion from two-dimensional image coordinates to actual spatial coordinates.When the information in the weld image space is converted to the actual space,only the two-dimensional coordinates of the weld in its actual space can be obtained.In order to obtain the three-dimensional coordinates,the light plane equation of the LSL needs to be solved,and the parameters of the light plane equation can be obtained by using the light plane(3)The experiments of the recognition of the right-angle fillet weld and the guiding of the welding robot.The experiment results show that,this system can obtain the characteristic points data of the right-angle weld joints with high efficiency,which can guide the welding robot to follow the coordinates of the weld joints to work.This system can be applied in actual industrial manufacturing for high quality intelligent welding.