Research On Trajectory Tracking Control Strategy Of Uncertain Manipulator Based On Iterative Learning

Among many control methods,Iterative Learning Control(ILC)is an effective research method for the working characteristics of the repetitive manipulator.However,due to non-periodic external disturbances and uncertain parameters,traditional ILC still has a lot of limitations,such as slow learning speed,low convergence accuracy,and poor control effect.Therefore,aiming at various disturbances and uncertain parameters in the trajectory tracking of the manipulator,this paper uses the combination of ILC and other intelligent control methods to address the above limitations encountered by the application of traditional ILC in the robot manipulator.The main work of this paper is described as follows:(1)Modeling of the manipulator.Using the D-H parameter method,this paper establishes the kinematic model of the two-arm robot Baxter through the Robotics Tools in Simulink,a MATLAB simulation tool.Based on the D-H parameter table and Baxter's kinematic model,a 3D visualization workspace of the Baxter robot is created using the Robotics Tools.The Euler-Lagrange method is introduced to set up the dynamic model of the robot manipulator,which provides theoretical support for the simulation experiment of the robot manipulator motion controller.(2)Two control schemes are proposed to track the position of a robot manipulator with some unknown dynamics,uncertain parameters and external disturbances when performing repetitive operations in task space.The first control scheme uses adaptive iterative learning control(AILC),which is similar to the feedback structure of PD,and combines adaptive control with iterative learning to design a track tracking controller.The second control scheme is composed of AILC and neural network.On the basis of the first scheme,RBF neural network is added.In the second control scheme,AILC is used to learn the uncertain parts of the periodicity of the manipulator,which are attributed to the repeated movements at the end of the manipulator.RBF neural network is taken to approximate and compensate for all nonperiodic uncertainties.The stability and convergence of the controllers for both schemes are strictly proved by the Lyapunov-based composite energy function.In the simulation,two control schemes are compared with those in classical and recent references.The results of simulation illustrate that the combination of AILC and neural network can better realize the position trajectory tracking of the manipulator,and improve the iterative learning speed as well as error convergence accuracy of the robot manipulator system.(3)A control scheme that combines a non-linear disturbance observer with AILC is presented to track the position of a robot manipulator with some unknown dynamics,uncertain parameters and external disturbances while performing repetitive operations in task space.First,the unmodeled dynamic model of the robot manipulator and the external disturbance are considered as a whole uncertainty.Then,the uncertainty is approximated and compensated by a non-linear disturbance observer,and the stability of the disturbance observer is verified by the Lyapunov function.Finally,an AILC is used to design the controller of the manipulator.The simulation results are verified by a two-link rigid manipulator.The simulation results demonstrate that the proposed control scheme can better track the position trajectory of the manipulator compared to the control scheme which is without nonlinear disturbance observer compensation.(4)An AILC scheme based on model global approximation is proposed to track position trajectory of a robot manipulator with completely unknown model dynamics,uncertain control parameters and various of external disturbances when performing repetitive operations in task space.The scheme first considers the dynamic model of the manipulator and the external disturbance as an uncertain item,then uses the neural network to approximate and compensate the uncertainties,and on this basis uses the adaptive iterative learning control as the controller of the manipulator.Finally,the controller parameters are adjusted in real time based on the Lyapunov function design control law.The simulation experiment is validated on a two-link rigid manipulator,and its results show that the proposed scheme can well track the position of the manipulator.