The Research Of Adaptive Control Algorithm Based On Tendon-driven Fingers

With the rapid development of the technology of robotic,the simple grippers as end of robot cannot meet the requirements of complex gripping tasks.In order to improve the flexibility of the robot hand,tendon-driven dexterous hands are used as end effector of robot,which can greatly improve the grasping ability of the robot.Therefore,tendon-driven dexterous hands have attracted the attention of researchers.Considering the friction between the tendons and the joints and the disturbance of the external environment,further research on the dexterous hand control algorithm is needed.This topic aims at the contact between the dexterous hand and the external environment and the disturbance of the external environment,and designs related control algorithms to improve the anti-disturbance of the control system.The specific research contents are as follows:(1)Aiming at the contact between the finger and the object and the external disturbance,an adaptive impedance controller based on BP neural network is designed.The neural network is combined with the impedance control to improve the contact force of the system when the tendon-driven fingers are in contact with the external environment.Adapt the ability and simulate the controller.The simulation results show that the control system has good stability when dexterous hands are in a complex and external disturbance environment.(2)Aiming at the problem that the controller parameters can not be adjusted in real time,an adaptive force controller based on deep neural network is designed.The neural network is combined with the traditional PID controller to design a force controller that can adjust parameters online and perform the controller.simulation.The simulation results show that when the dexterous hands comes into contact with the object,the control system can adaptively adjust the controller parameters according to the contact force error to achieve a good control effect.(3)A force/position hybrid control scheme based on fuzzy theory is designed.The interaction between dexterous hands and the environment is realized by controlling the tension and contact force of the dexterous hand tendon.The structure of the control system is elaborated and compared with conventional PID controller.The simulation results show that the fuzzy neural network controller has faster response than the conventional controller.