Research On Grasping Technology Of Industrial Robot Based On Machine Vision

The traditional way of grasping the workpiece by industrial robots mostly adopts the teaching programming mode,which has the disadvantages of poor flexibility,high requirement for placement position,and less precision and fault tolerance.In recent)years,with the increasing application of machine vision in industrial production,visual guidance for industrial robot grabbing operation has been paid more attention by researchers and industry,and has gradually become a hot spot of research.The introduction of vision has greatly expanded the breadth of robot grasping applications,enhanced the adaptability and reliability of the system.It is the trend of industrial automation development.At present,most of the workpieces in the research of industrial robot grasping based on vision are surface leveling,no hollow out,but in actual industrial production,the task of grasping the workpieces that it's sureface is hollow,not smooth and has a unfixed position occupies a large proportion in the daily industrial production.In view of the above problems,the research is carried out with the hollow workpiece like nut as the representative object.The following work has been completed:(1)an industrial robot grasping simulation system with visual guidance is built on the desktop manipulator,and the design of image acquisition subsystem,motion control subsystem and terminal fixture part are also given.The designed clamp adopts the inner and outer ring structure controlled by electric and pneumatic components,and can realize the reliable clamping of the hollow piece.(2)the coordinate transformation in the positioning of the workpiece under visual guidance is analyzed theoretically,the transformation relation of different coordinate systems is ascertained,and the camera calibration is completed.The post calibration error of the design system is less than 0.2mm;(3)on the basis of geometric template matching,feature point matching and the image segmentation based on traditional pulse coupled neural network,an improved pulse coupled neural network model is proposed.The experiment of high noise motor cover workpiece image is carried out.On the 640 x 480 pixel image,the segmentation and location of the workpiece can be realized in 0.5s.The calculation speed is improved by 200%than traditional PCNN' s under the same precision.(4)through the form of Uart bus,the interconnection of the image positioning system and the manipulator motion control system is realized,and the American national instrument(National Instrument)is adopted.The machine vision kit of NI and the hybrid programming technology of LabVIEW and MATLAB are used to complete the software design of the whole system for realizing the grasping of the mechanical arm under the visual guidance,and carrying out the experiment in the actual simulation system.The experimental results show that the designed system can reliably and effectively realize the clamp of the hollowing,uneven surface and unfixed workpiece by robot.Aiming at the problem of image segmentation of hollowed workpieces under high noise background,MPCNN algorithm is proposed.In order to enhance the mutual excitation effect of feedback to the PCNN model,the linear model is used in the modulation equation of the MPCNN model to accelerate the convergence speed of the algorithm.In order to quickly form the neuron activation area under the mutual excitation mechanism,MPCNN simplifies the input connection between neurons;at the same time,the segmentation threshold function is used to enhance the robustness of the algorithm under the noise situation.The robustness of the underwater sound.Compared with the maximum variance method and the iterative method,it is proved that the improved algorithm has the best segmentation effect and the shortest use time.The proposed MPCNN algorithm is robust and efficient.For sovle the issue of grasping the workpiece,the paper also designs a kind of inner and outer ring clamping structure,it realizes grasping and transferring by controling the cylinder,motor and micro screw of clamping structure.For the existing two finger fixture,it has better adaptability and can improve the efficiency of grasping.The overall clamping scheme proposed in this paper can effectively achieve hollowing workpiece clamping.The results of this study will be helpful to improve the theoretical research and engineering implementation of robotic arm grasping based on vision guidance.