Research On Machine Vision Workpiece Recognition And Positioning For Robot Assembly

Industrial robots have been widely used in the industrial field.However,when robots grab workpieces for assembly tasks,due to the uncertainty of the type and posture of the workpieces,it has brought difficulties to the grasping of industrial robots,and the accuracy of grasping Usually cannot meet the requirements of assembly positioning accuracy.In response to the above problems,this thesis combines machine vision systems with industrial robots,studies robot vision system models and calibrations,workpiece recognition and parameter optimization,robot grasping pose calculation and pose correction to improve the intelligence and accuracy of industrial robots' assembly tasks.The main research contents of this article are as follows:(1)For the robot assembly task,a binocular recognition and positioning system and a monocular pose correction system were constructed.The models and calibration processes of monocular and binocular vision systems are analyzed.The focus is on the research of hand-eye calibration in the form of " Eye-to-Hand ".The binocular system uses three-point hand-eye calibration and a new two-step hand-eye calibration algorithm which is simple,fast,and accurate for monocular systems is proposed.(2)For multi-model workpiece recognition problems,support vector machines are used to solve the problem based on the shape feature parameter set.A genetic particle swarm optimization algorithm is proposed for the feature selection and parameter optimization of the support vector machines.The accuracy of recognition after optimization reached 100%.(3)Aiming at the problem of workpiece positioning,the detection algorithm of corner and round hole features in the workpiece image is studied.The binocular vision system is used for reconstruction and positioning.A monocular vision system is used for accurate positioning after the workpiece is grasped,and a monocular posture correction algorithm is proposed to enable the workpiece with errors to be accurately installed at the theoretical position.(4)Set up the hardware system of the workpiece identification and positioning experiment platform and program the corresponding software system,and verify the validity and accuracy of the identification and positioning algorithm proposed in this paper through experiments.