Research On Flexible Positioning Technology Of Parts Based On Industrial Robot

The fusion of industrial robots and 3D vision has been widely used in the field of automation,which has become a major topic in the field of manufacturing automation in China.In order to enable a more complex operation of the robot,the robot need not only have a better control system but also a stronger environmental perception capability.A visually guided robot can achieve a better flexible operation of the robot.At the beginning of the paper,the hardware structure,software organization,system design scheme and operation process of the vision-based assembly positioning system are summarized.The end-effector design and stereo vision positioning system of the system are studied later,including:1.Design of Array End ActuatorIn order to realize the flexible positioning,grasping and assembling of the parts,an end-effector with array actuator is designed for the small parts with the characteristics of brittleness,curved surface and different specifications.In this paper,an end-effector design scheme is proposed,through hierarchical clustering of parts with similar characteristics,a special end-effector is designed for each type of parts,and finally the layout optimization method of end-effector array actuator is studied by finite element simulation.2.Multi-camera structure and calibrationAiming at the problems existing in the measurement work,the eye-in-hand robot vision system was constructed by using four cameras,and the hardware structure layout of the multi-camera system was guided by studying the influence of the visual system structure parameters on the measurement accuracy.Aiming at the problems of small public field of view between cameras and occlusion of field of view existing in multi-camera vision system,a multi-camera calibration method based on self-recognition marker is proposed.3.Feature extraction and pose calculationThe measurement object of the vision system is a circular reflection marking point that is affixed to the part.In this paper,the ellipse fitting method adopts the two-step strategy,the first step uses the improved Hough transformation to achieve the rough fitting of ellipse,the second step based on the least squares idea to achieve the precise fitting of ellipse.Based on the prior knowledge of the part and the measurement results of the visual system,the positioning of industrial robots is transformed into a 3d-3d pose estimation problem.In this paper,BA optimization algorithm is used to obtain the optimal solution.