Grasping Planning Of Multi-fingered Dexterous Hand Based On The Flexible Pneumatic Actuator FPA

With the rapid development of science and technology and the wide application of artificial intelligence technology,intelligent robots are exerting a subtle influence on people's life and work.As the medium of interaction between the robot and its surrounding environment,end-effector has attracted extensive attention.The multi-fingered dexterous hand simulates the structure and function of man's hand,which has overcome the disadvantages of conventional end-effector,such as simple structure,fewer degree of freedom and poor adaptability.Its good dexterity and adaptability play a positive role in promoting the application of robots.At the same time,reasonable grasp planning helps the dexterous hand grasp objects of different shapes flexibly and stably,which has great research significance.And robots have been extended to medical rehabilitation,service,entertainment and agriculture,among which the new dexterous hand shows great application value.This thesis is the grasp planning study of multi-fingered dexterous hand based on the flexible pneumatic actuator FPA.The main research work is as follows:(1)This thesis has given a brief introduction of the working principle of flexible actuator(FPA)and dynamic characteristics of bending joint.3-DOF anthropomorphic finger with three joints is proposed based on FPA.To achieve a high degree of modularity,five identical fingers and a palm constitutes the body structure of the multi-fingered dexterous hand.The structural parameters of the fingers and the layout program of the hand are determined by the bionic optimization method.Considering some aspects,such as,the target object model and grasping tasks,a general grasp planning program of multi-fingered dexterous hand is preliminarily formulated.(2)Machine vision is used to model the target object for a fuller picture of grasped object.Under the same condition,a set of common objects is selected for image collection.By image processing method,some parameters of target object in grasp planning,such as,geometry and basic posture,are obtained.(3)Based on the characteristic parameters of the target object and the grasping requirements,the thesis makes a research on the realization process of grasp planning of multi-fingered dexterous hand.The knowledge base of grasping mode is established and the planning of grasping mode is carried on according to the existing information.Using BP neural network,which is optimized with L-M algorithm,the grasping mode planner is designed.The simulation results show that the grasping mode planner has a feature of good classification.In the presence of internal grasping force,with friction cone constraints,the thesis establishes the contact force planning based on the stability from the perspective of geometry,finds the optimal converging point of the internal grasping force,and computes the corresponding internal grasping force and operation force,combined with the force closed triangle.(4)Using D-H method,the kinematics equation of multi-fingered dexterous hand is established.Based on the kinematics,experiments of finger position tracking are carried out.The response time of position tracking is less than 0.5s and the maximum error of angle tracking is 0.34 degrees.Then,the final grasp planning scheme is put forward,and based on the hardware such as developed sensor system,an experimental platform is set up for a group of objects' grasping operation.The experimental results show that the dexterous hand can grasp the target objects stably with grasp planning.