Model Predictive Uncalibrated Visual Servoing Scheme for Underwater Vehicle Manipulator Systems

This work presents a model predictive uncalibrated visual servoing scheme for an underwater vehicle manipulator system (UVMS). Visual information can be used to improve efficiency and accuracy in UVMS intervention tasks, traditional underwater visual servoing schemes suffer from uncertainties of UVMS kinematics parameters and camera calibration results, besides, systems physical constraints are seldom considered. In this work, a model predictive visual servoing scheme is used to deal with constraints of system states, Broyden’s method with recursive least square is utilized to approximate the composite Jacobian matrix without prior knowledge of system parameters, which is used to predict future states of the UVMS. Simulations are carried out to testify the effectiveness of the proposed model predictive uncalibrated visual servoing scheme.