Structural Design And Experimental Research Of Bionic Dexterous Hand For Fruit Picking

Agriculture is the foundation of our national economy.It is an important industry to ensure economic development and social stability.With the continuous development of intelligent technology,improving the level of agricultural mechanization and intelligence has become an important research topic.However,the existing picking robots are not yet able to meet the market demand regarding picking efficiency,flexibility and versatility.Although some picking robots use soft robots as the end-effector,their dexterity still has a big gap compared with human hands.In order to improve the dexterity and versatility of the end-effector to be compatible with grasping objects of different sizes and shapes,a bionic dexterous hand for fruit picking is proposed in this paper to address the problems faced by current picking robots.The main research work is as follows.Inspired by the movement of the thumb and index finger in human hand picking work,this paper developed research on the design and fabrication of a two-finger bionic dexterous hand with movable fingertips and palm.Inspired by the origami structure,a zig-zag structure of the finger’s inner core was designed,and a parametric optimization study was conducted with the help of FEM.This paper explored the grasping performance of the bionic dexterous hand in the envelope grasp,finger pinch mode and movable fingertip motion mode through theoretical modeling combined with experimental methods.The empirical relationship between the grasping object size and the optimal palm angle in the envelope grasp mode was constructed,and the optimal grasping strategy in this mode was obtained.The experimental results show that the safe working range of palm angle in the envelope grasping mode is30°～120°,which can grasp spherical fruits with diameters of 70 mm～152 mm and weights less than 15.0 N.The safe working range of palm angle in the gripping mode is 0°～30°,which can grasp spherical fruits with diameters of 0～70 mm and weights less than 7.9 N.The movable fingertip motion mode can achieve efficient grasping of thin sliced objects.In order to explore the fruit picking performance of the bionic dexterous hand,a picking robot system is built in this paper.The target detection module of the system uses the Detectron2 target detection framework and Mask-RCNN algorithm,and the mean average accuracy of the algorithm is evaluated to be 96.04%,which can meet the detection requirements of the project.The motion control module was built based on the ROS platform.Finally,the bionic dexterous hand was mounted on the six-axis robotic arm to implement fruit picking experiments.The results show that the success rate of apple picking by the picking robot with the bionic dexterity hand is 80% and the average time required to pick an apple is 8.97 s.