An Underwater Vehicle-Manipulator System(UVMS) can be applied to diverse underwater works such as a construction, inspection and maintenance for ocean structures. It can be also used to pick up and carry an object using a underwater-manipulator for autonomous manipulation in the water. Remotely Operated Vehicles(ROV), which are physically linked with a surface ship through a tether cable, are mostly utilized as a vehicle platform of the UVMS. However, it is difficult to operate a ROV among complicate structures due to a tether cable. They also have some disadvantages that it is hard to actively cope with an environment change such as ocean currents. Autonomous Underwater Vehicles(AUV) that can carry out various works actively and autonomously are considered to overcome some problems of the ROV and used as the vehicle platform of UVMS in this study. A hovering-type AUV equipped with multiple thrusters should maintain the specified position and orientation in order to perform given tasks. This system is usually called Dynamic Positioning(DP) system. Besides, the simple control allocation algorithm based on a scaling factor is presented in this study. Moreover, it is necessary for the robust control algorithm to conduct successfully given missions in spite of a model uncertainty and a disturbance. In this paper, the robust DP control algorithm based on a sliding mode theory is addressed to guarantee the stability and better performance despite the model uncertainty and disturbance. Next, if the trajectory of the vehicle is not given in advance, the whole system has a redundancy and it has an infinite number of solution in an inverse kinematics to satisfy the given trajectory of an end-effector. Therefore, when some specified tasks are performed, the redundancy resolution is introduced to determine the desired joint trajectories of the manipulator. The redundant vehicle-manipulator system can have various combinations of joint velocities that do not affect the given velocity profile of an end-effect and this may induce a self-motion of a vehicle. The performance index based on the concept of Zero Moment Point(ZMP) is proposed to enhance the dynamic stability of the whole system. In order to generate the joint angle trajectories of the manipulator, a redundancy resolution is performed to minimize the distance between the position of ZMP and the mass center of the underwater vehicle. Finally, a series of simulations and experiments are conducted to verify the availability of the generated trajectories and performance of the designed robust controller.