Fingertip stiffness control using antagonistic pairs of polyarticular tendons drive system

This paper proposes a tendon driven system with antagonistic pairs of polyarticular tendons to simulate a variable stiffness of a human fingertip. The stiffness of a robot finger should be adjusted according to a situation that includes large position errors or collisions, because it tolerates errors of object positions and dimensions in assembly task and absorbs collision energy in hitting or catching task. The variable stiffness therefore contributes to stable grasping and dexterous manipulation. Besides, a stiffness of a human fingertip can be controlled by consciously contracting antagonist muscles when it does not contact anything such as in a preshaping phase. This study simplifies a tendon driven system of a human finger and then searches feasible parameter settings of the tendon driven system with antagonistic pairs of polyarticular tendons through numerical simulations. Finally it is experimentally confirmed that the proposed tendon driven system has similar stiffness to a human finger.