Optimizing Study Of Tomato Picking Manipulator

The picking of tomato in agricultural production is mainly done by hand at present.This method requires high labor demand and labor intensity,but the picking work is inefficient.In addition,due to the complexity of the tomato picking environment and its vulnerable characteristics,the mechanization and automation of tomato picking has become a major research challenge and hotspot.In this context,this paper takes tomato as the research object,and innovatively designs a tomato picking device to achieve nondamaged picking of tomato fruit.The main research contents and conclusions of this paper are as follows:（1）In order to improve the versatility of the device,this paper selects the larger strawberry tomatoes and the smaller peanut tomatoes “DanDong-409” as the research objects and conducts a study on their physical and mechanical characteristics.The geometric parameters of the tomatoes and the static friction coefficient of the fruit surface are measured,mainly including the average weight,maximum transverse width,longitudinal width,geometric mean diameter,sphericity,and the average static friction coefficient between the tomato fruit and the silicone and rubber materials for both large and small tomato fruits.Compression experiments are carried out on two varieties of tomato fruit at different maturity levels using the TA.XTC+TA.touch texture instrument to obtain the compression load-deformation law and the relationship between fruit compression resistance and maturity level for large-fruited strawberry tomatoes and small-fruited peanut tomatoes based on different maturity levels.The experimental results show that the minimum positive stress of tomato skin damage is15.75 N,and the minimum stable clamping force of the end effector is 2.61 N.（2）Based on the above experimental results,a tomato picking end effector based on bionic idea is designed and prototyped according to the characteristics of the tomato fruit and the needs of the picking environment.This end effector consists of a shell,a finger gripping module,a holder rotation module and a cutting module.The end effector machine（STM32F407 microcontroller as the core chip）controls a set of drive servos to realize the picking action of the end effector.The fingers consist of a U-shaped finger and a wide finger with a curved finger skeleton and a flexible material on the finger surface,with a thin film pressure sensor underneath.The purpose is to enable the fingers to be adaptively attached to the surface of the fruit to achieve a stable grip while reducing damage to fruits and vegetables.The U-shaped finger and the wide finger are analyzed statically in ANSYS Workbench to check the strength and stiffness of both fingers.The maximum stress on both fingers appears at the root of the finger.The maximum stress on the U-shaped finger is 89.15 Mpa,and the wide finger is 103.53 Mpa,both below the yield strength of the material at the root of the finger,meeting the design requirements.（3）Generally,it is necessary to judge the maturity of the fruits in advance when picking process is taken,so as to realize the accurate grasp of the tomato fruits that meet the picking requirements.In this paper,based on the deep learning method,the recognition research of tomato fruit based on machine vision and maturity is carried out.A tomato dataset is made.Images of tomato fruits with different maturity levels are acquired and the dataset is expanded and enriched by image mirroring,rotation,cropping,Gaussian noise interference and contrast processing.After the images are labeled,they are converted into txt type files for training,and the dataset is divided.Using the PyTorch deep learning framework and based on the python assembly language,the Yolov5s-based tomato detection algorithm implementation environment is built.The trained model has a good recognition effect with a mAP of 97.1%,which meets the requirements for accurate fruit recognition by the visual part of picking robot in a natural picking environment.（4）The motion control test is carried out on the end effector.After 50 tests,it is concluded that the upper computer（main controller）and the lower computer（end effector controller）can achieve good signal communication.When the end effector receives the corresponding command,it can respond quickly and meet the working requirements of the picking operation.In addition,in order to explore a more suitable material for the finger surface of the end effector,this paper applies ANSYS to simulate the gripping test of the tomato picking end effector,using rubber and silicone as the finger surface of the end effector to grip the tomato fruit.The test results show that when using the fingers of the silicone finger surface to hold a large-fruited tomato,the stress on the tomato is mainly distributed on the skin of the tomato and the inner part of the side of the wide finger.The maximum stress appears on the tomato skin on the side of the wide finger,and the maximum stress is about 0.61 Mpa.The maximum stress when holding a small-fruited tomato appears on the tomato skin on the side of the wide finger,and the maximum stress is 1.08 Mpa,compared with rubber material as the finger surface（the maximum stress for the large one is 0.74 Mpa,and the maximum stress for the small one is 1.14Mpa）,the end effector fingers with silicone material as the finger surface are more suitable for grasping tomato fruits.Under laboratory conditions,the end effector prototype is used to grasp tomatoes with a success rate of 86% for largefruited tomatoes and 81% for small-fruited tomatoes.Tomatoes of different maturity levels are grasped with almost no damage.Therefore,the end effector designed in this paper can achieve stable and non-damaged grasping of tomatoes of different maturity.