This thesis is aimed at improving the robot grasping with unknown disturbance. The whole grasping control is divided into three stages: first, generating a stable contact force for robot hand to grasp and manipulate an object with unknown disturbance; second, finding an optimized solution and distributing the resultant contact force to all contact fingers; and third, applying the optimized fingertip forces to a hand force controller with tactile sensor feedback. The integrated hand-arm system is robust to the manipulating process, and can be applied to the daily life task. Experiment of fastening a screw is conducted and the result is satisfactory.