Design Of Robotic Manipulator System And Research On Its Adaptive Neural Network Control

Since its emergence,the robotic manipulator has achieved considerable development and extensive research,and a variety of mechanical manipulator s have been derived.The mechanical manipulator has been widely used in the human production and life,greatly improving the level of human production and life.With the development of computer technology and artificial intelligence technology,the development of robot technology has entered a new stage.The ultimate goal of robot research is to develop intelligent robots similar to humans.Therefore,the research on robotic manipulator with human manipulator characteristics is an important direction for future manipulator research.Based on the SRU3 research of the robot service center of the University of Electronic Science and technology of China,this paper designs and studies the robot manipulator system with human manipulator characteristics.The aim of this paper is to design the manipulator system and to study its control algorithms.Firstly,the system platform of the manipulator is designed.According to the characteristics of the human manipulator structure,the manipulator is designed to be a 3 joint and 6 degree of freedom rigid manipulator.This structure enables the designed manipulator to simulate the motion characteristics of the human manipulator as far as possible.The control system of the manipulator is designed as 3 layers.The upper level control adopts PC to solve the control algorithm of the manipulator,and the MCU is used for information processing and motor driving,and the lower level is the driving motor and sensor device.The whole control system uses a distributed control structure,and each drive motor on the manipulator is controlled by a MCU.The control system software architecture is constructed based on the COS-III real-time operating system,in order to improve the real-time and stability of the whole system.Then,based on the designed structure of the manipulator,the model of the manipulator is built.In this paper,the forward and inverse kinematic model and the dynamics model of the manipulator are established respectively.The kinematic model is modeled by the screw theory based modeling method,and the dynamic model is built based on the modeling method based on the screw theory and Lagrange equation.By using the screw theory to establish the kinematics model of the manipulator can be organic rigid transformation and screw motion together,and the kinematics model described as production of exponentials formula(POE).Compared with the commonly used D-H method this modeling process is simpler,and can avoid the drawbacks of D-H method.In general,the inverse kinematics model of the manipulator is solved by the Paden-Kahan sub problems.In this paper,a geometric method is used to solve the problem,and its operation process is simpler and more intuitive.Finally,the model based control algorithm and adaptive control algorithms are designed,considering the control requirements under the known system model and unknown model.Adaptive algorithm used in adaptive controllers is based on radial basis function neural network(RBF),and uses its universal approximation of neural network to approximate the manipulator system.At the same time,solve the control under the actual conditions,we consider the complete known state of the system and some unknown conditions.Finally,the Lyapunov stability principle and the MATLAB simulations are used to analyze the effectiveness of the controllers and verify the stabilities of the system under different controllers.Finally,with the support of the manipulator system of the Baxter robot,two adaptive neural network controllers are experimentally verified with the manipulator system,which provides a further proof for the effectiveness of the algorithm.