This paper presents a study on autopilot design for an unmanned surface vehicle subject to dynamical uncertainty, time-varying ocean disturbances and unmeasured yaw rate. An output feedback adaptive steering law is developed based on a state observer and a neural network using iterative updating law. As a result, this approach is able to achieve automatic yaw control in presence of dynamical uncertainty and time-varying ocean disturbances without yaw rate measurement. Using a Lyapunov-Krasovskii functional, it is proven that the error signals are uniformly ultimately bounded. Simulation result is given to show the effectiveness of the proposed method.
L INTRODUCTION
RECENT years have witnessed a surge of interest in developing efficient methods for autopilots design of marine surface vehicles. Advent of sophisticated controllers for autopilot design contributes for the vehicle to carry out tasks with reduced manpower, adequate economy, sufficient reliability, and optimum performance [1], There has been a number of results in the recent decades focused on autopilot design. In [2], a fuzzy PID autopilot is proposed for surface vessels, In [3], a sliding mode con troller based on simple sliding surface is developed for ship autopilot In [4], a robust adaptive backstepping controller for autopilot control of ship with parameter uncertainty is reported. In [5], the adaptive fuzzy control and dynamic surface control method are combined to perform the tasks of autopilot for marine vehicles, In [6], a ship autopilot control system which consists of the bang-bang controller and the fuzzy neural controller is developed, In [7], a neural network (NN) based adaptive autopilot for marine applications is proposed. In [8], a direct adaptive NN control algorithm is designed for a class of ship course autopilot with input saturation, Note that most autopilot design are developed by assuming that the yaw and yaw rate are all measurable [2]-[8], How ever, an autopilot controller without yaw rate measurement is much preferred in practice in the sense that it reduces the implementation cost In addition, derivative-based NN scheme have been used for autopilot design [6]-[8], while the NN weights are all assumed to be constant However, they may not be able to capture rapidly varying disturbances or require the use of unrealistically high adaptation gain. This paper reports an autopilot design for a robotic un manned surface vehicle subject to dynamical uncertainty, time-varying ocean disturbances and unmeasured yaw rate. The control architecture is developed by integrating a state observer and an NN using iterative updating law. The state observer is employed in place of a reference model, and the estimate states follow both the reference model and the true states. By using the iterative updating law, the proposed approach is able to capture rapidly varying disturbances and improve the transient behavior without increasing the effective adaptation gain [9]. The stability analysis using a Lyapunov-Krasovskii functional shows that all er...
