Hierarchical image-based visual serving of underwater vehicle manipulator systems based on model predictive control and active disturbance rejection control

The image-based visual servoing (IBVS) problem of a redundant underwater vehicle manipulator system (UVMS) with an eye-in-hand camera is investigated in this paper. With consideration of system constraints, dynamic uncertainties, and the absence of vehicle velocity sensors for a typical UVMS, we propose a hierarchical control architecture, which is composed of an unscented Kalman filtering (UKF)-based vehicle motion estimator, a kinematic model predictive IBVS controller, and two decoupled dynamic velocity controllers for the vehicle and the manipulator, respectively. In details, the underwater vehicle velocities are estimated by UKF with visual measurements based on the UVMS kinematic model. With these estimates, a nonlinear model predictive controller (MPC) is designed to solve the IBVS problem of UVMSs and generate velocity commands for the underwater vehicle and the manipulator, simultaneously. The visual kinematic model of UVMSs is used to predict the future trajectories, and the field-of-view, manipulator joint angle, and UVMS velocity constraints are handled when solving the optimization problem. To reduce the system complexity of UVMSs, the high-dimensional UVMS’s dynamic model is decoupled into two low-dimensional models of the vehicle and the manipulator, and a dynamic inversion-based active disturbance rejection control (DI-ADRC) method is developed for dynamic controller design. The physical interaction effects between the vehicle and manipulator are considered as external disturbances acting on each other, which are estimated by extended state observers (ESO) in ADRC. The simulation experiments with a typical UVMS are performed to demonstrate the effectiveness of the proposed hierarchical IBVS controller in dealing with system redundancy, constraints and uncertainties.