Robust adaptive formation control of underactuated autonomous surface vehicles with uncertain dynamics

In this study, a neural network-based control design is developed for underactuated autonomous surface vehicles moving in a leader-follower formation, in the presence of uncertain leader dynamics as well as uncertain local dynamics caused by coriolis and centripetal force, hydrodynamic damping, unmodelled hydrodynamics and disturbances from environment. By online approximating the uncertain leader dynamics, the proposed method only uses the measurements of line-of-sight range and angle by local sensors, no other information about leader is required for control implementation. The proposed control law is model-independent that does not rely on the accurate model of the vehicle, which is difficult to obtain in practice. Based on the Lyapunov synthesis, it is proved that with the developed neural formation controller, all the tracking errors converge to a small neighbourhood of zero. Simulation results demonstrate the effectiveness of the method.