Design and control of under-actuated tendon-driven robotic hand reflecting human hand motion analysis

Abstract

We propose a novel design framework for the under-actuated tendon-driven (UATD) robotic finger. In this study, we decompose the system equation based on its stiffness to separate the actuated space and the un-actuated space. The actuated space appears during the free motion when no disturbance is exerted, whereas the un-actuated space is observed when there is contact. We measure the index finger joint angles and contact force during both the free motion and the contact, and search for the efficient design parameters (stiffness, active tendon routing, initial configuration) to mimic the human finger motion. We formulate the optimization problem to find those parameters and we consider the system constraints, s.t. finger thickness and unilaterality of tendon-driven mechanism, simultaneously. We also design a simple PD-based control to track the desired joint angle, and we show the simulation results.