Predictor LOS-based trajectory linearization control for path following of underactuated unmanned surface vehicle with input saturation.

This paper investigates a novel robust path following control scheme for unmanned surface vehicle (USV) with unknown dynamics, currents and actuator saturation. In the guidance loop, an improved predictor line-of-sight (PLOS) guidance law is presented, which is used in any parametric paths. In the control loop, we propose a yaw rate controller and a surge speed controller using trajectory linearization control (TLC) technology, finite-time nonlinear tracking differentiator (FTNTD), minimal learning parameter (MLP) technique and auxiliary system. The advantages of the developed scheme are that, first, the proposed PLOS provides the estimates of unknown sideslip angle and currents; second, the introduced TLC has a concise structure and strong robustness, and enhanced TLC only needs two parameters to be adjusted. A MLP technique is constructed to approximate lumped unknown dynamics, where the norm of all the weights is estimated instead of estimating each element to reduce computational burden. A FTNTD is concurrently used to achieve satisfactory differential and filter performance. Subsequently, a smooth auxiliary system is applied to handle input saturation constraint on actuator. Theoretical analysis illuminates that the system is semi-globally uniformly ultimately bounded (SGUBB). The effectiveness of the developed scheme is verified by simulation comparison and the error quantification performance. • A fuzzy predictor line-of-sight (FPLOS) guidance law is developed to calculate the desired heading angle and identify the unknown sideslip angle and ocean currents. • An enhanced trajectory linearization control (TLC) technology is first used to design a path following controller based on three-degrees-of-freedom model for underactuated USV, which only needs two parameters to be adjusted. • A finite-time nonlinear tracking differentiator (FTNTD) is concurrently introduced to achieve satisfactory differential and filter performance. • A minimal learning parameter (MLP) technique is constructed to approximate lumped unknown dynamics, where the adaptive law uses single parameter instead of all weights online learning to reduce computational burdens. • An auxiliary dynamic system that is governed by smooth switching function is developed to compensate for the saturation constraint on actuator. [ABSTRACT FROM AUTHOR]