I. Introduction
Powered prosthetic hands have been successful and of great benefit to amputees. However, there have been many problems with powered prostheses in the past [1] that have yet to be fully resolved by systems that are currently available [2]. Due to high gear ratios and rugged mechanisms, there is a large amount of friction and backlash in prosthetic hands. High friction effects cause a substantial deadband in the motor of the hand where a large input signal must be given before the hand begins to move. Regardless of the source of the input to the prosthetic hand, large effective inertia due to large gear ratios causes overshoot when transitioning from velocity to force control when gripping an object. This drastically reduces the ability of the user to carefully grasp delicate objects. The problem of force overshoot when initially grasping an object is also hampered by the dichotomy of force–velocity control for hand prostheses. The dual nature of controlling prosthetic hands is that very rapid changes in position and velocity are typically desired until the hand closes sufficiently to grasp the object. Then, very small changes in position and velocity are desired to adjust the magnitude of the applied grip force, depending upon the stiffness of the hand and grasped object. Because the fingers of prosthetic hands are made of rigid materials, there is no easy way to control the manipulator's mechanical impedance to improve grasp force sensitivity as well as in natural systems [3].