Multi-stage cubature information filter based nonlinear model predictive scheme for steering control of an autonomous underwater vehicle under sensor/actuator failure

Abstract

The hydrodynamics of an autonomous underwater vehicle (AUV) is an extremely intricate and multidimensional problem. The dynamic interaction between orientation change and the consequent shift in hydrodynamic forces exerts substantial effects on AUV’s stabilization and steering efficacy. Furthermore, a malfunctioning sensor/actuator leads to unexpected outcomes when executing steering maneuvers. A nonlinear model predictive control (NMPC) scheme incorporated with the observer was proposed in this work for the AUV to perform steering maneuvers. A two-stage high-degree cubature information filter is aimed at accurately tracking sensor/actuator faultiness, undetermined hydrodynamical parameters, and uncertain perturbations. Using the observed AUV state/parameters, a predictive control strategy for the anti-disturbance model has been devised. This research has extensively examined multiple types of real-world situations involving the impact of ocean currents, parametric sensitivity, and repercussions of sensor/actuator faults. A variety of indices, including mean square deviation (improvement of 5.36%) and root mean square error (improvement of 6.29%), are assessed using the proposed control framework and contrasted with the standard form of nonlinear model predictive controller to identify its efficacy and acceptance on the depth tracking scheme.