Neuro-fuzzy control of underwater robot based on disturbance compensation

Adaptive nonlinear output feedback control of GD-FNN(generalized dynamic fuzzy neural network) for underwater robot motion control using forming filter for wave disturbance is presented. This method completely construct nonlinear and uncertain parts of underwater robot by online adaptive learning algorithm without knowing fuzzy neural structure and training phase in advance. Output feedback control guarantees initial stability of system. The speed that controller needs is got by dynamic compensation and Kalman filter, which solve coupling problem between freedom. Closed-loop stability of system is proved by Lyapunov and structure of controller is simple without accurate mathematical model of system. Finally, an underwater robot model is adopted to verify the effectiveness of the method.