Research On Development And Control Method Of A Pneumatic Continuum Robotic Arm

Pneumatic continuum robot is a robot that has no rigid joints and is composed of compliant materials.It has extremely high flexibility and maneuverability,and it can be used in complex structural environments well.What's more,it's quite safe for humancomputer interaction.The robotic arm developed in this thesis is a kind of pneumatic continuum robot.As an actuator,it can be applied to the tasks like operation assistance,grasping and handling and so on in many fields,such as industry,agriculture,daily life and nursing.In this thesis,the research on development and control method of the pneumatic continuum robotic arm is carried out,a prototype of the pneumatic continuum robotic arm is designed and produced,and the pose accuracy of the movement of the robotic arm gets improved.The bellows are the basic unit of the pneumatic continuum robotic arm.In this thesis,the design of the bellows is based on finite element calculation and analysis.The bellows of different cross-sections and different tapers are compared in their elongation,bending performance and tangential stiffness.The appropriate parameters are selected to design and manufacture the bellows.And the design and assembly of the robot arm body is finally completed.At the same time,the pneumatic system of the pneumatic continuum manipulator is designed based on the rapidity and flexibility of the manipulator's movement and the safety of human-computer interaction.For continuum robotic arm,it is difficult to extend to multi-segment,highdimensional manipulators because of the complexity of kinematic modeling based on segmented constant curvature.Therefore,this thesis collects the movement data of the robotic arm to train the neural network to complete the inverse kinematics mapping,which makes the inverse kinematics calculation of the manipulator easier and more convenient.However,the neural network fitting inverse kinematics mapping still has a large positioning error,and the robustness to load changes is not good.This thesis proposes a control strategy combining neural network inverse kinematics and iterative learning to converge the manipulator pose error,and finally completes the control system design of the robotic arm.In the end,a large number of experimental researches on the robotic arm are carried out on the experimental platform.The motion performance experiment tests the positioning accuracy,repeat positioning accuracy and movement speed of the robotic arm.The error convergence experiment proves that the control strategy of neural network inverse kinematics and iterative learning can effectively converge the pose error under no-load and load conditions.The operation experiment enables the pneumatic continuum robotic arm to complete the corresponding tasks well in the actual application operation scene,which reflects the stability and reliability of the robotic arm.