This paper treats the control system of the underwater manipulator considering its working environment. We construct the system of the underwater manipulator using Learning Feed-Forward Controller, which is adaptable to the changes of the dynamics of the manipulator. The system controls position and force hybridly. Then the experiments of the system were carried out, it is shown that the proposed control system has good performance. Finally, we propose a trajectory planning method to save the energy consumption to control the manipulator.
Recently, ROY and AUY are often used in the maintenance or" offshore structures and the development of ocean resources. These operations are carried out with the underwater manipulator. This manipulator system is consists of trajectory planning part and control system. Trajectory planning part decides path of the end effector of the manipulator such as movement of the manipulator from the initial position to the final position. Then by the control system the manipulator is controlled to trace the desired trajectory precisely. There are two problems when these operations carried out. The first problem is that precise control of the manipulator is difficult by ordinary feedback control method, because the dynamics of the multi-link type manipulator has nonlinear characteristic and unknown parameters like friction of the joint and water resistance acting on the manipulator arm. The second is that the energy of the untethered type ocean robot is limited, since the robot cannot receive energy from the mother ship. So we consider the control system of the manipulator as Fig.1. In the trajectory planning part, we propose the method saving energy consumption to control the position of the manipulator. This method can generate the trajectory in real time using the neural network and/approximation by the polynomial.