Design And Research Of Chinese Wolfberry Picking Robot System Based On Binocular Vision

Automated production and processing is the inevitable trend of agricultural development in China.The Chinese government has already established the strategy of revitalizing agricultUre through science and education.It will take the popularization of automation equipment,the improvement of production efficiency and the reduction of labor cost as the direction of future agricultural production development.The picking of fresh fruits of Chinese wolfberry has always been a difficult problem hindering the development of Chinese wolfberiy industry.The picking of fresh fruits of Chinese wolfberry mainly depends on manual work.The low picking efficiency and the high cost of picking have become the bottleneck restricting the development of Chinese wolfberry.With the rapid development of social economy and the rapid aging of population in China,it is an urgent problem to realize the automatic picking of Chinese wolfberry fruit.Based on the above reasons,it is of great significance to develop an eflficient and low-loss Chinese wolfberry picking robot based on machine vision.In this paper,a picking robot based on binocular vision is designed.It consists of a 6-DOF manipulator and a binocular stereo vision system.The specific research contents of this paper are as follows:A binocular stereo vision system was built,and camera calibration method was studied.Zhang Zhengyou method was selected to complete the internal parameters calibration of binocular camera and the relative position calibration of left and right cameras through the software of MATLAB.The internal parameters matrix and relative position matrix of left and right cameras were obtained.The image segmentation methods of Chinese wolfberry fruits were studied.By analyzing the characteristics of Chinese wolfberry fruits in RGB and HSV color space,samples of Chinese wolfberry fruits and background were collected.Two image segmentation models,BP neural network and SVM,were established.The image segmentation of Chinese wolfberry fruits was completed by using BP neural network and SVM combined Laplace operator and corrosion-expansion morphology.An improved least squares method and back-propagating(BP)neural network optimized by genetic algorithm are proposed for hand-eye system calibration.The BP neural network calibration model was established by using the coordinates of the target in the camera coordinate system and the robot arm coordinate system as the input and output.The results showed that compared with the improved least squares method and BP neural network model,The BP neural network model optimized by genetic algorithm can significantly reduce the calibration error and speed up the convergence.The average errors for the measured distances between the tested results and the predicted ones by the proposed genetic BP neural network model,the conventional BP neural network model,and the improved least squares method were 1.7126mm,2.4002mm,and 3.3771mm,respectively.Taking IRB-1200 six-axis manipulator as the research object,the linkage coordinate system of the manipulator is established by using the improved D-H method,and its kinematics model is established and analyzed.The forward and inverse kinematics solutions of the manipulator are obtained.