Saturated kinetic control of autonomous surface vehicles based on neural networks

Abstract

This paper investigates the saturated kinetic control of autonomous surface vehicles subject to unknown kinetics and limited control torques. The unknown kinetics stems from parametric model uncertainty, unmodelled hydrodynamics, and environmental forces due to wind, waves and ocean currents. By approximating the unknown kinetics using neural networks, a bounded kinetic control law is proposed based on a saturated function, with the main advantage being that the control input is known as a priori. The resulting closed-loop kinetic control system is proved to be input-to-state stable.