| The extraction and planning of robot machining trajectory is the core technology of intelligent robot teaching,which directly affects the speed and accuracy of robot machining system.Traditional teaching methods for industrial robots are point-to-point,where the robot moves repeatedly according to fixed points and requires the robot operator to calibrate the machining points and write teaching programs.Although traditional teaching methods have been widely used in industry,when the type or size of the workpiece changes,it requires operators to recalibrate the point position,therefore,the automation and intelligence of traditional teaching methods are comparatively low.This paper addresses the above-mentioned problems for two application situations: welding and gluing,by using the line laser sensor to model the workpiece in 3D and achieve the automatic extraction and planning of robot machining trajectories.The specific work includes the following aspects:(1)We designed and constructed a 3D vision-based trajectory extraction system.The hardware part of the system includes robotic arm,sensor,PC and conveyor belt.An improved hand-eye calibration algorithm based on a stepped calibrator is proposed,which has a simple calibration process and better accuracy than the traditional standard sphere calibration method.The software part of the system includes the design and development of the PC software,which integrates common point cloud processing algorithms and can realize line laser point cloud pre-processing,processing trajectory extraction,real-time display of point cloud data and intelligent trajectory planning of the robot.(2)In order to address the problems of the point cloud processing algorithm such as low real-time and easily disturbed by noise,in this paper,combining the advantages of line laser sensor contour can be separated,we propose a maximum distance-based line laser contour key point extraction algorithm,which uses the maximum distance method to search for key points on the contour data of the laser contour sensor,and then transform these key points to the three-dimensional space,the experiment results show the algorithm is better than the traditional point cloud key point extraction algorithm in terms of time and accuracy.(3)In order to solve the problem of poor machining quality due to more position deviations when loading the workpiece,a robot machining trajectory correction algorithm based on point cloud registration is proposed,and the point cloud registration algorithm is used to calculate the transformation matrix of the workpiece position change by aligning the workpiece point clouds scanned twice,and the teaching points are recalculated using this transformation matrix to achieve the purpose of correcting the trajectory of the robot arm.The experiment shows that the algorithm accurately corrects the robot trajectory within a certain range based on the point cloud registration.(4)A line laser point cloud registration algorithm that fuses contour features is proposed to address the problems of long registration time and low registration accuracy of ICP and Fast ICP registration algorithms.By searching the feature points of the line laser contour lines and using these feature points for the iteration of the alignment algorithm,the improved registration algorithm reduces the number of iterations and requires less initial positions of the source and target point clouds.The experimental comparison of ICP and Fast ICP point cloud registration algorithms shows that the improved point cloud registration algorithm outperforms the traditional point cloud registration algorithms ICP and Fast ICP in terms of speed and registration success rate when applied to the line laser point cloud.In summary,this paper proposes a 3D vision-based processing trajectory planning method that can reduce the number of robot teachings and improve production efficiency in actual production,and it has wide application prospects in the field of automated welding and gluing. |
PDF Full Text Request