Quasi-Newton method based finite-time fuzzy tracking control for uncertain unmanned surface vehicles: Indirect and direct approaches

Accurate trajectory tracking for unmanned surface vehicles (USVs) is critical for various marine applications but remains challenging due to inherent system uncertainties, complex hydrodynamic disturbances, and the need for real-time adaptation. By constructing a fuzzy logic system (FLS)-based scheme, conventional adaptive fuzzy control methods face a trade-off between convergence accuracy and computation efficiency. To address this, a novel Quasi-Newton method based FLS construction framework is put forward, by which the second-order convergence of learning parameters is ensured while simultaneously avoiding the prohibitive cost of computing exact Hessian matrices. In this paper, the design dependence between the hybrid USV uncertainties and the output of FLS is characterized through a predefined cost function. Then, an indirect and a direct Quasi-Newton method based fuzzy learning control approaches are elaborately illustrated. The finite-time theory is incorporated into the control design process, by which the semi-global practical finite-time stability (SGPFS) of tracking error signals via the indirect and direct control approaches is demonstrated. Finally, simulation examples and the related comparison results are conducted to show the efficacy of the developed Quasi-Newton method based fuzzy learning control approaches in driving the USV.