Three Dimensional Spatial Information Modeling Of Complex Welds In Welded Robots Based On Laser Structured Light Scanning

The robot's intelligent welding is the main trend of the development of automated welding technology.At present,most of the practical applications at home and abroad of the welding robot path planning for the manual teaching.Such robots have extremely strict requirements on the consistency of the welding environment.Pre-setting the welding path and related process parameters does not have the adaptability to the errors in machining and assembling of the actual welding,which leads to the deviation between the welding seam and the teaching trajectory.At the same time,the precision and efficiency of teaching depend on the artificial visual accuracy and proficiency,so it is difficult to meet the requirements of modern welding with high efficiency,high quality and high stability,thus limiting its application in many fields.The application of visual sensor enables robots to acquire the welding path autonomously,overcomes the influence of uncertain factors in the trajectory planning process on teaching efficiency and welding quality,enhances the intelligent level of robots,and has high application value.Also solved the teaching of programming inefficient,unstable,greatly improving the programming efficiency.In this paper,a set of autonomous planning system of laser welding based on laser structure is developed.The algorithm is transplanted and the acquisition of the original data points of large and complex welds is realized by means of multi-step scanning with variable pose.Calibration is a bridge between the two-dimensional information in the coordinate system of the sensor and the three-dimensional information in the space coordinate system.A robot hand calibration method based on linear constraint is proposed for robots with end-line-structured optical sensors.The optimization model was constructed based on the linear constraint.The improved Powell algorithm and the penalty function method were used to solve the optimization model.The experimental results show that the calibration accuracy of this method is within 0.2mm,which improves the calibration precision effectively.The method requires only a piece of relatively straight edge calibration template,without precise calibration target,the calibration process is simple and fast,suitable for on-site calibration.Aiming at the problems of poor timeliness of 3D data processing,this paper proposes to generate gray-scale images from 3D data and propose a complete set of image processing algorithms.Mainly including median filter,ostu algorithm to obtain threshold generated binary map,based on connected domain area threshold filtering,swelling based on smoothing process,canny operator extraction weld edge and zhang-suen algorithm refinement of the edge process.After obtaining the pixel coordinates of weld feature points,it is proposed to separate the edge of the workpiece separately and extract the pixel coordinates of the edge of the workpiece by using image algorithm.The welding seam information is reconstructed three-dimensionally.The advantages and disadvantages of the cubic B-spline curves and the DP algorithm are compared.The DP algorithm is used to establish the welding seam position model and compress the data points to remove the redundant data.Establish the model of different types of welded joints,correct the coordinates of the characteristic points and obtain the geometric parameters of the welding seam;then fit the edge of the workpiece and the end of the welding seam based on the least square method,find the starting point and the ending point of the welding seam.Using weld proximity vector and forward vector establish the pose model of welding seam and torch.According to the method of this paper,three kinds of welding seam,such as plane straight line,flat arc curve and space curve,are experimented to obtain the starting point of welding,geometric parameters of joints and the three-dimensional information of welding seam.Experimental results show that the accuracy of linear weld is higher than that of curved weld,the accuracy of flat weld is higher than that of curved weld,and the maximum error is less than 1.5mm.In terms of geometric parameters of joint,the error of height direction is less than 0.5mm,The error in width direction is less than 1mm.The experiment results of many ways in three-dimensional information of weld seam show that the error between calculated point and actual weld point is less than 1.5mm.The system can basically meet the robot's actual welding production requirements.