Robust Predictive Visual Servoing of USVs With Wave Perturbations Considering FOV Constraint

Abstract

For unmanned surface vehicles (USVs) under wave perturbations, this article proposes a two-phase robust visual servoing method that maintains the image features within the field-of-view (FOV). First, a visual trajectory planning approach is proposed based on hybrid A*. The node expansion and Reeds–Shepp curve searching strategies in hybrid A* are improved to ensure the visual trajectory satisfies the FOV and nonholonomic constraints. Moreover, a minimum acceleration optimization method is integrated in hybrid A* to consider the USV dynamics limitations. Then, a robust nonlinear model predictive control (NMPC) scheme is proposed to track the visual trajectory under wave perturbations. For wave induced uncertainties on the roll, pitch and heave degrees of the USV, a feature correction method is adopted. For the uncertainties on the surge and yaw degrees, a tightened state constraint is proposed to tackle all the admissible realizations of the uncertainties in the NMPC scheme. Sufficient conditions are provided to guarantee the robust recursive feasibility of the NMPC algorithm. Additionally, it is shown via theoretical analysis, simulation results, and experimental results that the tracking errors will robustly stay in a small set around the origin, and the USV can finally arrive at the goal position within a finite time.