Research On Weld Seam Extraction And Trajectory Planning Based On Machine Vision

The production of equipment such as containers and large ships requires a huge amount of welding components to be spliced.The welding quality directly affects the quality and safety of the entire equipment.Existing component welding mostly uses manual welding and robot teaching welding.Aiming at the shortcomings of manual welding method,such as high labor intensity,low welding efficiency,and poor adaptability of robot teaching welding method,the need to re-teach the position of the workpiece,and high labor cost,this thesis proposes a machine vision based automatic extraction and welding automatic generation method of seam track.This thesis mainly researched from the following aspects:(1)Vision system calibration.Based on the imaging principle,t he robot coordinate system,image coordinate system and world coordinate system are analyzed;the single target of the camera is prepared for the acquisition of the weld seam image;the distortion correction of the camera reduces the image radial and tangential directions distortion error improves the accuracy of welding seam extraction;the use of robot hand-eye calibration improves the response speed of the welding end of the robot,and creates conditions for the improvement of the accuracy of the movement trace of the welding end of the robot.(2)Weld seam image recognition algorithm.In order to reduce the influence of redundant noise inside and outside the component on edge detection,a component edge extraction algorithm based on adaptive Canny edge detection and maximum contour points is proposed.Adopting adaptive median filtering for edge preservation and denoising;by adding 45 and 135 gradient templates to calculate the gradient amplitude,the Canny operator is further enhanced to extract detailed edge information and anti-noise ability;using the minimum error method,the threshold self-adapted to the setting,a clearer weld edge was extracted;for the internal and external noise of the component con tour is not easy to remove,a maximum contour point denoising method was proposed;the welding segment of the component was extracted by the hough transform,preparing for the welding path planning.(3)Kinematic analysis of six-axis welding robot.Using the D-H method,the robot kinematics model is established,the robot motion form is discussed,and an inverse kinematics solution method for robot pose separation is deduced.The Carlo sampling method is applied to obtain the robot motion space;necessary preparations are made for the robot motion control and welding path generation in the welding section.(4)Welding segment coordinate conversion and trajectory planning.The extracted feature points are reconstructed in space points and converted into robo t coordinates;for the T-shaped speed planner,the speed transition is not smooth,there is impact and the standard S-shaped speed curve is not adaptable,and an efficient and stable universal S-shaped speed is constructed planner;and apply the obtained general S-shaped speed planner to robot interpolation.In the end,the relatively clear and accurate welding seam was extracted through Matlab simulation and the smooth and efficient movement trajectory of the robot welding seam was obtained,which was of great significance to the actual factory welding.