Two-DOF Manipulator Trajectory Tracking Control Based On Unfalsified Control

The research of manipulator control is a comprehensive and rapid developing discipline. It has been highly appreciated by industry and academia. Kinematics and path planning of manipulator is an important research content in robotics. As its inherent coupling, uncertain disturbance and other factors, lots of difficulties are found in establishing an accurate mathematical model. Therefore, it is difficult to deal with such issues based on modern control theory. Considering the above problems, this thesis is to design an unfalsified controller which is not based on models.The unfalsified control strategy designs a controller set that consists of a performance objective and measured input-output online data without plant models or assumptions on the plant. The theory works by eliminating candidate controllers that do not satisfy cost function by evolving experimental data. The controller which remained to the last is a controller which satisfies a performance specification of cost function. Main work is as follows:Firstly, basic PID unfalsified control method is presented. PID unfalsified controller is designed and applied to the manipulator trajectory tracking control system. Accurate tracking control of manipulator is achieved by selecting the control parameters of PID properly.Secondly, considering the disadvantage of finite candidate controllers set, gradient unfalsified control method is presented. This thesis designs a gradient unfalsified controller, which expands the set of candidate controllers from finite one to infinite one. The simulation results prove that gradient unfalsified control scheme is effective.Finally, variable step-size gradient unfalsified control method is introduced in this thesis. The selection of the step-size is based on inexact one-dimensional search technique. The simulation results prove the effectiveness and robustness of the proposed method.In order to verify the effectiveness of the three algorithms, simulation and experiments are performed by Matlab/Simulink. Simulation results of manipulator verify the validity of the control strategies and their advantages in tracking performance. The advanced gradient unfalsified control method increases the tracking speed with high precision. Besides, the control system is robust and it can resist the impact of certain disturbance.