Learning-Based Neural Adaptive Anti-Coupling Control for a Class of Robots Under Input and Structural Coupled Uncertainties

This paper investigates the learning - based adaptive anti-coupling control issue for the robots under input and structural coupled uncertainties. In this paper, the input and structural coupled uncertainties are modeled and transformed into a system state related term, an internal state related term and a system input related term. With the aid of the actual exponential input-state stability and the dynamic auxiliary signal, the internal state related uncertainties can be suppressed. By utilizing the properties of the robot dynamics and several special nonlinear functions, the system state related uncertainties can be handled. Moreover, to overcome the system input related uncertainties, an indirect control law and the adaptive boundary estimation law have been designed. To simplify the control structure, the neural networks have been introduced as online approximators. Finally, a novel learning-based intelligent adaptive anti-coupling control structure has been established for the robots. The simulation results revealed the satisfactory control performance of the proposed anti-coupling control algorithm.