Trajectory Tracking Control Based On Sliding Mode Variable Structure For A 5-DOF Upper-limb Exoskeleton

Currently,the research of exoskeleton has become a main area of robotics through the whole world due to its significant value in military,industry,agriculture and healthcare,such as enhancing the power of the human limb,helping the wearer finish the exhausting work more easily and providing effective rehabilitation training for the disabled.Exoskeleton system is highly nonlinear,strong coupling and uncertain.Furthermore,we have very high requirements for comfort and security of the exoskeleton,which brings more difficulties to the design of its control system.In this thesis,trajectory tracking control of exoskeleton with nonlinearity,uncertainties,external disturbances and unavailable states is mainly studied.The main contents are as follows:Firstly,with the assumption that the exoskeleton is a series rigid body,dynamics model of a 5-DOF upper-limb exoskeleton containing shoulder,elbow and wrist joints is established based on Lagrange method,and the detailed parameters of the system are presented.The composition and characteristics of each part of the model and its influence on the system are analyzed,as well as several characteristics of the exoskeleton such as nonlinearity,strong coupling,uncertainty,disturbance and so on.Secondly,a robust controller based on RBF neural network and sliding mode control is designed considering the nonlinear uncertainties in the 5-DOF exoskeleton.Considering the defect of sliding mode method in leading to severe chattering,the neural network is used to estimate the unknown nonlinear function of the system and compensate to the controller.The chattering can be greatly reduced since smaller gain of the sliding mode control is needed.The weights of RBF neural network are adjusted online by adaptive algorithms and the adaptive rate is derived from Lyapunov method.Thus,the phase of complicated off-line training is avoided while the stability of the system is guaranteed.Thirdly,the finite time disturbance observer is introduced to estimate the system disturbances,and the terminal sliding mode control method is applied to realize the finite time control of the exoskeleton under external disturbance.The lumped disturbances made up of the system uncertainty and external disturbance can be accurately estimated within a limited time by the terminal sliding mode disturbance observer.Then the controller is designed based on the fast nonsingular terminal sliding mode.Based on the above results,the singularity in terminal sliding mode is avoided and the convergence is speeded up,thus the tracking error converges to zero in finite time.Finally,considering the situation with no measurement of angular velocity,the feedback controller is developed only the joint angle available.The lumped disturbances of the system is treated as an expansion state,and a finite time extended state observer is proposed to observe the state and lumped disturbances of the system simultaneously.Then a sliding mode controller is designed based on the observed value of joint angular velocity and the lumped disturbance,so that the limited time tracking of command signals under unmeasured angular velocity is realized.