DRFNN adaptive observer based sliding mode tracking control of an underactuated surface vehicle

This paper address the problem of trajectory tracking control of an USV based on nonlinear adaptive observer using dynamic recurrent fuzzy neural network (DRFNN). In order to control an underactuated surface vehicle (USV) efficiently, knowledge about the position, velocity and attitude of the USV is needed. For low-cost USV, the sensor suit can only provide measurements of position and yaw. The proposed observer can estimates the unknown nonlinear terms in the USV's dynamics without exact knowledge about parameter of Coriolis- centripetal matrix and nonlinear damping matrix. The adaptive observer scheme is proved to be uniformly ultimately bounded. Furthermore, a new sliding manifold definition with exponentially stable combination of the conventional manifold and it derivation is presented. The new sliding model control algorithm guarantees the fast convergence rate and stability in tracking as well as robust against observer's estimation error. Simulation results demonstrate the effectiveness of the proposed approaches.