Research On Identification And Positioning Technology Of Parallel Tea Picking Robot

Tea picking is a vital part in the tea industry.There are two main ways of tea picking working at present: traditional picking and mechanical picking.Traditional picking,which is also called manual picking,has a greater demand for labor and will lead to problems such as labor shortage in the tea picking season.The existing mechanical picking schemes mostly use various tea picking machines for picking,which requires human involvement as well.Automation and selective picking can not be achieved by above methods.With the continuous development of sensor technology and robotics.The researches of intelligent picking robot with selectivity have attracted more and more attention of relevant scholars.Among these researches,the key technologies of picking robot such as structure design,target recognition and positioning have become the research focus.Aiming at tea bud as the target of identification and positioning,this paper proposed a structure scheme of picking robot based on Delta parallel structure,in order to optimize the efficiency and quality of tea picking work,and improve the identification and positioning accuracy.The research content of this paper mainly includes the following three aspects:(1)Modeling and Simulation analysis of Delta parallel robot.Firstly,the structure of Delta parallel robot is modeled by Solid Works 3d modeling software,and the parameters of each main mechanism are determined.Secondly,D-H parameter method and analytical method are used to model the forward and inverse kinematics of Delta parallel robot.Then in order to analyze the speed and acceleration conversion relationship between the control motor and the end-effector,Jacobian matrix is established based on the modeling results of forward and inverse kinematics.Finally this paper simulates and analyzes the workspace size,working path,speed,acceleration and other parameters of the structure end-effector based on co-simulation of Matlab and Adams simulation software.By analyzing the simulation results,the correctness of the forward and inverse kinematics model based on D-H parameter method and analytic method can be verified.Meanwhile,the actual movement of the workspace to meet the needs of the picking work,which provides a theoretical basis for the design of positioning system based on vision technology and a reference value for the follow-up researches.(2)Research of tea bud recognition based on deep learning.In order to achieve selective picking and location of tea buds,it is necessary to identify tea buds.With the development of computer technology and deep learning,detection algorithms for small objects have also been developed.In this paper,the improved YOLO-v4 deep learning algorithm is used for research.The development and characteristics of YOLO series algorithms are analyzed.Meanwhile,the structure and improvement strategy of YOLO-v4 algorithm are described as well.Finally,the YOLO-v4 deep learning network is trained based on the environment of Open CV and VS2015,and the experiments related to tea identification are completed then.The experimental results show that the improved YOLO-v4 deep learning network can complete the identification of tea buds,but it also has some problems like low recognition rate.Therefore,this paper uses the optimization algorithms of the improved data set augmentation strategy to optimize the original data set.And the improved YOLO-v4 model is used for training to each strategy above.Then tea identification has been carried out respectively.The experimental results show that the improved algorithm is helpful to improve the identification rate and accuracy of tea buds in the improved YOLO-v4 neural network model,which lays a good foundation for the subsequent spatial positioning work.(3)Hand-eye calibration and ranging based on binocular vision principle.After the completion of the identification work,it is also necessary to conduct spatial positioning and ranging of tea buds to guide the picking manipulator to reach the target position of tea leaves.This paper adopts binocular camera for hand-eye calibration and ranging experiment.In order to complete hand-eye calibration,eye-to-hand system has been analyzed and the internal and external parameters of binocular camera are calibrated by Zhang calibration method based on Matlab.Finally,through the principle of triangle ranging,ranging experiment for using the center point of recognition frames.Experiments results show that the error is within reasonable acceptable range.And it is verified that the system can guide Delta parallel tea picking robot to complete tea picking and positioning work.