Research On Configuration Design And Grasping Control Of Humanoid Dexterous Hand For Fruit And Vegetable Picking

At present,fruit and vegetable picking is still mainly completed by manual,which has low efficiency and high production cost,so it is urgent to realize the automation of fruit and vegetable picking.The key of automatic fruit and vegetable picking operation lies in the research of fruit and vegetable picking device,which mainly has two difficulties:first,because the shape of fruit and vegetable is not standard,the traditional industrial manipulator can not form a good package for it,and it can not be used to grab small fruits and vegetables because of its large size;Second,because the surface of fruits and vegetables is generally fragile,the traditional picking device cannot control the output force,which is easy to cause damage to fruits and vegetables in the picking process.Therefore,the design of human-like dexterous hand that can realize various picking actions and grasping force control can be used as a breakthrough point in the research of fruit and vegetable picking devices.Taking humanoid dexterous hand as the research object,this paper carried out the research on the configuration design and grasping control of humanoid dexterous hand for fruit and vegetable picking from three aspects:driving mechanism,configuration parameters and intelligent perception.The main research contents are as follows:Based on the application requirements of humanoid dexterous hand,the overall system scheme of dexterous hand is designed.The structure and function characteristics of human hand are analyzed,and the underactuation principle is applied to design the transmission mode of double tendon rope and the overall structure of human-like dexterous hand.From the perspective of integrated control system design,the hardware system of humanoid dexterous hand is designed by combining the force output and force perception function of humanoid dexterous hand.The upper computer software is developed to realize the human-computer interaction of dexterous hand.The configuration design method of humanoid dexterous hand is studied and its optimal configuration parameters are determined.The maneuverability index to measure the dexterity of thumb manipulation and the form-closed index to measure the grasping stability of humanoid dexterity hand were proposed.The optimal configuration parameters of humanoid dexterity hand were determined by particle swarm optimization algorithm.The rationality of configuration parameters was verified by motion space simulation.Build a prototype dexterous hand.Aiming at the sliding and grasping force instability which may occur in the grasping process of humanoid dexterous hand,the flexible grasping control method of humanoid dexterous hand is studied.The initial grasping expected force of humanoid dexterous hand was determined by hand grasping experiment.Based on the Cartesian impedance space,the output force tracking controller of humanoid dexterous hand is designed by combining nonlinear tracking differentiator and proportional integral differential control.Based on the Haar wavelet transform,the vibration information of the sliding object grabbed by the humanoid dexterous hand was detected by its detail coefficient,and the grasping force control method was designed by combining the output force tracking controller and the sliding detection method.In order to realize the humanoid dexterous hand using different grasping actions for different shapes of fruits and vegetables,the automatic generation method of humanoid dexterous hand grasping posture was studied.A directed box processing method was designed to describe the three-dimensional characteristics of the object to be grabbed.Based on a variety of common fruit and vegetable generation data sets in daily life,a back-propagation neural network was constructed which could be used to generate the grabbing posture.