Grasping and handling delicate objects with robotic systems using force closure method needs a sophisticated control capable of providing stable grasping of the object with minimum possible normal force. This work presents a new controller iired by human method of grasping. The proposed controller only needs measurement of the contact forces, instead of measuring relative velocity and acceleration of object. To estimate friction coefficient in uncertain situations, a mathematically simple algorithm is also presented which also does not need measurement of relative motion. It is shown that in noise-free situation, the proposed estimation algorithm guarantees robustness of the controller against variations in the friction coefficient. To hold and manipulate the object with two multi-link fingers, we need a control scheme capable of controlling force and motion in similar direction. Most of position-force control algorithms proposed so far divide the work space of manipulators into two suace and control position in one suace and force in the other one. Meanwhile, for safe grasping purpose we should apply force at minimum necessary level which itself is not a known quantity during motion, so the desired value of normal force is not known forehand. This work propose a new position-force algorithm and employ that to accomplish pinching and moving delicate object with two multi-link fingers. Several numerical examples are provided to show versatility of the proposed method. Key Words Safe Grasping, Multi-Phase Friction, Coulomb Friction, Friction coefficient estimation, Dexterous Hand, Manipulation Object and Controller.
