Research On Adaptive Control Of Robotic Manipulator Systems With Uncertain Parameters

As the robotic manipulators are used in more common scenarios,complex working environments and long-time of heavy working,it will inevitably be affected by complex and unknown uncertainties,and the operation performance of the end-effector of the manipulator will be significantly reduced.Furthermore,with the increasingly demanding industrial applications in our country,in addition to requiring high stability and highprecision performance indicators for the robotic manipulator systems,the finite-time convergence of control errors,the low calculation cost of the controller,and the simplicity of the operating mode of the robotic manipulator are also required.Although the adaptive control of manipulators with parameter uncertainties has always been the hot spot by domestic and foreign scholars,the aforementioned problems are still longstanding,yet unsolved control problems in the robotics field.Aiming at the uncertainty of the abovementioned robotic manipulator systems and the neglected control performance index,this paper mainly completes the following research contents:(1)For the manipulator system,on the basis of considering the uncertainty of dynamics and kinematics,the perturbation of actuator parameters is also considered.By designing a separate actuator parameter compensation adaptive law,the excessive actuator parameter compensation errors and unknown disturbances are eliminated effectively.Then,by using inverse Jacobian matrix technique,a set of adaptive control algorithm with the separation property of kinematics and dynamics is proposed,and the stability of the system is proved strictly by Lyapunov stability theory.Therefore,the adaptive tracking control problem of the manipulator system with multiple uncertainties is solved,and the simple operation mode of the manipulator was realized only need to design the joint reference velocity instruction.(2)A manipulator system with multiple parameter uncertainties is considered,the sliding mode observer and neural approximator methods are introduced to solve the uncertain kinematics and the uncertain dynamics including unknown torque and external interference,respectively.After that,combining with the low-cost neural network adaptive mechanism,the n*N(n is the degree of freedom of the manipulator,N is the number of neural network nodes)is reduced to one adaptive law,and an adaptive control algorithm for manipulator systems with multiple-uncertainties based on low cost neural approximators is proposed.In addition,the strict system stability analysis and proof are given by Lyapunov stability analysis method.The proposed method solves the low-cost adaptive tracking control problem of the considered system,and the control accuracy and robustness of the controller are further improved.(3)The finite time convergence problem of manipulator systems with uncertain kinematics,uncertain dynamics and unknown disturbances is studied.A low-cost neural adaptive finite-time tracking control method that can track the target trajectory in a finitetime is proposed.First,the sliding mode observer is used to deal with the uncertain kinematics to ensure that it accurately estimates the position of the end-effector in the task-space within a finite time.Then,a neural approximator is used to estimate the uncertain kinematics and unknown external disturbances,and a new sliding mode manifold switching function is designed to solve the singularity problem.Based on these strategies,a finite-time adaptive control technique based on low-cost neural approximator is proposed,and the Lyapunov stability and finite-time stability analysis methods are adopted,which prove that all closed-loop signals are bounded,and the tracking error converges to an arbitrary small neighborhood of the origin within a finite time.(4)An adaptive control algorithm based on low-cost neural network is proposed to deal with the manipulator system with uncertain dynamics.The effectiveness of the control algorithm is verified by strict stability analysis and digital simulation experiments.Furthermore,the low-cost neural network adaptive algorithm is applied to the Sawyer manipulator platform with 7 degree-of-freedoms(DOFs),which verifies the practical application value of the algorithm,and the problems of high-DOFs manipulator system controller computational burden and the difficulty of neural network control in the practical control task are solved.