Research On Adaptive Fuzzy Control Of Robotic System

Uncertainties are ubiquitous in robotic systems,such as,inaccuracy in measurement,modeling,friction,load changes,and external disturbances.Among various control methods,adaptive fuzzy control is an effective way to obtain high performance control effects.However,the traditional adaptive fuzzy control still has some limitations,such as low learning ability,poor control effect and slow convergence time.Therefore,this thesis studies the adaptive fuzzy control of robotic systems,and systematically solves the above-mentioned limitations encountered in the application of traditional adaptive fuzzy control methods.The main contributions of this paper are summarized as follows:(1)For the position trajectory tracking of a single robot with unknown nonlinear dynamics,an adaptive fuzzy sliding mode control scheme is proposed.The switching function in sliding mode control is used as the input.The controller is designed according to the approximation ability of the fuzzy system.The adaptive law was designed based on the Lyapunov method to adjust the parameters required by the controller in real time.The simulation results show that the adaptive fuzzy sliding mode control can achieve better position trajectory tracking and effectively reduce the chattering compared to the traditional sliding mode control.(2)For the consensus problem of nonlinear dual-robot system with multiple-input and multiple-output,an adaptive fuzzy wavelet network control is proposed.A consensus controller is designed to ensure that the two robots are synchronous in the joint position and joint velocity,respectively.Then the fuzzy wavelet network compensator is designed combining the approximation ability of fuzzy system and the learning ability of wavelet network to deal with the unknown nonlinear dynamics in the system while improving the approximation ability and the approaching speed of the system.Finally,the adaptive laws are designed based on the Lyapunov function to adjust the parameters of the controller in real time.The performance of the proposed control is compared to that of the traditional adaptive fuzzy control in the experiments where the experimental setup is established by two Phantom Omni robots.The results show that the proposed adaptive fuzzy wavelet network control can achieve better consensus of the joint position and joint velocity while effectively reducing the adjustment time.(3)For the position tracking problem of the dual-robot system with time delays,an adaptive fuzzy wavelet network control scheme based on sliding mode is proposed.The integral sliding mode is designed as the inputs of the controller to reduce the number of system inputs and fuzzy rules.Then,for the problems of time delays and unknown nonlinear dynamics of the system,the fuzzy wavelet network controller is designed combined with theapproximation ability of the fuzzy system and the learning ability of the wavelet network to improve the approximation ability and the approaching speed of the system.Furthermore,the switching control is designed to reduce the approximation error.Finally,the adaptive laws are designed based on the Lyapunov function to adjust the parameters of the controller in real time.The performance of the proposed control is compared to that of the traditional adaptive fuzzy control in the experiments where the experimental setup is established by two Phantom Omni robots.The results show that the proposed adaptive fuzzy wavelet network sliding mode control can achieve better tracking of joint position and joint velocity and effectively reduce the effects of time delays.(4)For the position tracking of multiple robotic systems,an adaptive fuzzy wavelet network-based control scheme is proposed.The joint positions,joint velocities and control torques of the robots are selected as the input of the fuzzy functions.The controller is designed combining the approximation ability of fuzzy system and the learning ability of wavelet network.Meanwhile,the adaptive law is designed based on the Lyapunov function to adjust the parameters of the controller in real time.The performance of the adaptive fuzzy wavelet network control is compared to that of the traditional adaptive fuzzy control in the simulations.The results show that the proposed control scheme can achieve better trajectory tracking and effectively reduce the adjustment time for multiple robotic systems.