Research On Seam Track Measuring System Based On Stripe Type Laser Sensor

The application of industrial robots in the field of welding has been developed rapidly,and the coverage is more and more extensive.The "teaching and reproduction" model is the most common working mode in robot welding operation.To ensure that this working mode can be applied accurately to the environment dominated by welding operations,the operator needs to finish spot welding in the process before welding,so as to facilitate the accurate location of the workpiece.Because the workpiece itself has the shape and position deviation,as well as other unavoidable factors such as stamping deformation and burr and so on will bring welding edge positioning error,so that the actual welding track and the teaching track does not coincide,there is tracking deviation problem,reduce the welding accuracy,and ultimately not meet the welding quality requirements.In general,correcting welding trajectory is a prerequisite to improve the quality of welding workpiece.Therefore,this paper proposed that the welding trajectory of the workpiece was measured and corrected before welding.A stripe laser sensor was used as a detection device.This paper designed a weld measurement system based on laser stripe sensor.The laser sensor was installed on the end of the welding robot.The sensor might detect the weld line and obtain the cross-section two-dimensional geometric information of the weld and extract the feature points.All the feature points were fitted to obtain a complete weld trajectory using NURBS curve.This research comes from the national science and technology major projects "five thousand units with completely independent intellectual property rights for machine tool automation and industrial robot development project"(Item No.2015ZX04005006)and the Guangdong province science and technology major projects "for CNC machine tools and robots integrated technology"(Item No.2014B090920001).It is about how to measure and correct the welding track of the workpiece before welding,and discuss the related problems.The main contents of this paper included the following parts:Studied the data processing method of the laser data,which includes two processes: filtering and smoothing.The method of filtering was Hampel filtering algorithm,which could identify the outliers in the original data and eliminate them.The smoothing method used a first-order weighted linear least square method.The smooth function in MATLAB provided six different ways for smoothing.The example showed that the first order weighted linear least square method could be used to smooth the data,and the noise interference in the data could be greatly reduced without losing the data characteristics.Studied the hand-in-eye calibration method between stripe laser sensor and robot based on SDP.The identification model of the relationship between the laser measurement coordinate system and the robot coordinate system was established.The model was used to identify the model parameters and identify the physical feasibility based on the semidefinite programming method.The average error of the calibration result was-0.2863 mm.Considering the various errors in the measurement process,the results could meet the requirements of welding precision.Studied the welding path fitting algorithm based on NURBS curve.Researched on interpolation and approximation algorithm from the two aspects of local and global.Introduced four fitting algorithm of the global curve interpolation,local curve interpolation,global approximation and local approximation.By comparing the advantages and disadvantages of the four algorithms,the appropriate algorithm for fitting the curve of the weld is obtained.Designed and constructed an automatic welding platform for 8 degree of freedom robot.The hardware system included YASKAWA 6 DOF welding robot and positioner and LPS 36 HI type laser stripe sensor.The control system included PC software modules and YASKAWA DX200 controller.The platform was used to verify the above theoretical research.The experiment was divided into three aspects: the experiment of the filter and the feature point extraction algorithm for the collected laser data;experimental study of calibration algorithm for the conversion of laser sensor and robot end;fitting results and analysis of NURBS curve of weld feature points.Through the error analysis of the experimental results,it could be seen that the average error of the welded seam trajectory was 0.8mm,which meet the requirements of welding accuracyThe innovations of this paper are as follows:(1)Proposed a new method of weld track correction based on stripe laser sensor.A measurement system consisting of a laser sensor,an industrial robot,a position changing machine and an industrial control computer was set up.The sensor was fixed on the measuring range of the workpiece by the robot.The overall characteristics of the weld was measured by rotating the workpiece on the displacement machine.(2)Proposed the method of NURBS curve fitting based on the welding seam feature points.The laser data was filtered and smoothed.Extracted the feature points from the processed data.According to these feature points,the NURBS curve fitting was used to obtain the complete weld trace.(3)A semi definite programming based model identification and calibration algorithm is proposed.By analyzing the transformation relationship between the laser measurement coordinate system and the robot's terminal coordinate system,the identification model is established.The model was used to identify the model parameters and identify the physical feasibility based on the semidefinite programming method.The calibration results with high accuracy could be obtained under the condition of physical feasibility.The advantage of this method was that the calibration results could meet the physical feasibility of the model,that was,to satisfy the constraints of the attitude matrix orthogonality.The identification model had higher stability.