| Kiwi fruit picking is characterized by high labor intensity and tight picking time.At present,there is no commercialized picking equipment.The existing kiwi fruit automatic picking technology mainly has the problems of high damage rate of picker and high cost of serial manipulator equipment.In this study,taking the shed standard orchard as the picking environment,the kiwi fruit picking technology research and device design based on machine vision and parallel manipulator were completed through the design of fruit visual positioning system,finger plucking bionic picker and delta parallel manipulator.The main research contents and results are as follows:(1)Analysis of operation scenario and picking technical scheme.Firstly,the spatial characteristics and spatial distribution of kiwifruit orchard were investigated,and the technical scheme of bottom-up picking with inverted parallel manipulator was determined;By analyzing the characteristics of color tone and contour of kiwi fruit inverted image,the fruit recognition method based on color tone and the fruit location method based on binocular vision were determined;Finally,based on the single fruit size and multi fruit distribution characteristics of "Hayward" kiwi fruit,combined with the mechanical properties of mature fruit and manual picking action,the principle of finger plucker based on force control detection was determined.(2)Research on fruit recognition and location algorithm based on binocular vision.The binocular camera is used to take fruit photos,and the principle of tone recognition is used to judge the fruit maturity.The binocular positioning method based on sgbm stereo matching algorithm outputs the horizontal projection coordinates(x,y)and depth distance Z of mature single fruit in real time to realize the spatial positioning of single fruit.By analyzing the recognition characteristics of "Hayward" kiwi fruit,the kiwi fruit contour feature is approximated as an ellipse,and a fruit contour segmentation method based on color difference R-G automatic threshold segmentation is determined.Combined with the color characteristics of kiwi fruit calyx,the two-dimensional plane recognition of kiwi fruit calyx points is completed,and the recognition rate is 73.7%;Based on the binocular ranging principle,stereo matching was carried out on the left and right eye images,and the fruit spatial positioning algorithm based on sgbm stereo matching algorithm was determined.The experiment showed that the average positioning error was about 54 mm,and the error rate was about 12.3%,which basically met the use requirements.(3)The design of finger pulling bionic picker.Based on the physiological and biomechanical characteristics of the fruit stalk which is easy to cause the separation layer of the fruit pedicle to break under the radial force,a finger plucking bionic picker with controllable clamping force,separation of clamping and picking action and integration of picking and transportation was designed.Based on the analysis of the fruit structure size and spatial distribution characteristics of "Hayward" kiwi fruit,the design idea of "bottom-up,one fruit at a time,intelligent and light,and harvesting as a whole" was put forward.The fruit fixation method based on force control clamping,the approach detection method based on TOF micro proximity sensor and the separation method based on finger bionic structure design were determined,and a finger plucker imitating the hand picking action was built,The finite element analysis of the stress on the key part of the fruit pedicle finger was carried out,and the design scheme meeting the stress requirements was obtained after optimization.The final experiment showed that the success rate of picking was 78.0%,and the average picking time of single fruit was 5.53 s.(4)Optimization of control algorithm for multi fruit picking motion of parallel manipulator.The kinematics model of the parallel manipulator is built,and the solution from the end coordinates to the motor joint angle is realized;Based on PH curve,the picking path of the manipulator is optimized to avoid environmental obstacles while realizing fast and stable movement.According to the spatial characteristics of the kiwi orchard,considering the cost,performance,size and other factors of the manipulator,the development scheme of the picking manipulator based on the delta parallel manipulator is determined.Through three-dimensional simulation and testing,the key parameters of the manipulator size that meet the preset workspace are obtained;The kinematics model of the parallel manipulator is established.The base coordinate system with the motor axisymmetric center as the origin and the tool coordinate system with the end picker clamping center as the origin are established.The inverse kinematics relationship of coordinate system transformation is established.The manipulator motion planning algorithm of S-shaped trajectory combined with PH curve path is designed and verified by simulation.(5)Whole machine construction and picking effect verification.The control system is aimed at the automatic fruit picking operation in the scaffolding Kiwi orchard.The target fruit is identified and positioned by machine vision to realize the multi-objective sequencing of the fruit.Through the coordinate conversion between the camera and the manipulator,the picking path and trajectory of the delta manipulator are planned in real time to realize the coordinated control of the three subsystems: Fruit identification and positioning,bionic hand picker and delta manipulator.In the hardware of the whole machine control system,the control system architecture is designed,in which the upper computer is responsible for visual identification and positioning and picking point sequence output,and the lower computer is responsible for the driving of the manipulator and the control of the picker.In terms of control system software,visual system software,end picker system software and manipulator system software are compiled,and the whole machine system software is compiled by combining the picking strategy and communication architecture.A 1.8m high scaffolding Kiwi orchard was built in the laboratory environment to verify the picking effect.The experiment shows that the success rate of the developed device is about 61.2% when the car body is stationary,and the average picking efficiency of a single station is about 491.1 /h,which basically meets the expected requirements. |
