A robust model predictive path following controller for an Autonomous Underwater Vehicle

Abstract

A robust model predictive controller (RMPC) based on the H∞ criterion is proposed for accomplishing effective path following motion plan of an Autonomous Underwater Vehicle (AUV) in face of uncertainties, external disturbances, and actuator constraints. A factorized Extreme Learning Machine (FELM) dynamic model of an AUV is proposed in this paper to identify the dynamics of AUV. Further, the parameters of the FELM model are updated at every sampling instant to accommodate any parameter variations in the AUV dynamics. The aforesaid FELM dynamic model is used to design a robust model predictive controller using Linear Matrix Inequality. The simulation and experimental results envisage that a robust path following performance is achieved in face of uncertainties and disturbances whilst avoiding actuator saturation. A comparison in terms of tracking performance is carried out between the proposed controller and the H∞ state feedback controller. From the obtained simulation and experimental results, it is inferred that, in face of actuator constraints and external disturbances, the proposed controller demonstrates efficient tracking performance compared to the H∞ state feedback controller.