Precision Control of Fingertip Force by a Biorealistic Hand with a Pair of Neuromorphic Muscles.

Abstract

In this paper, the ability of precision control of fingertip forces was investigated in an antagonistic cable-driven prosthetic hand with neuromorphic twin of muscles. Surface electromyography (sEMG) signals collected from able-bodied subjects' forearm were processed and used as alpha motor commands to drive the neuromorphic muscle models. A pair of antagonistic muscles were cascaded by two servo motors to control the index finger. Force control performance was tested by pressing a spring with a fixed stiffness using the fingertip, where forces with varying target levels were regulated with visual feedback. Two able-bodied subjects performed the precision force control task with the prosthetic index finger by sEMG signals and the intact hand. One subject was tested with force level changes of 0.1N, and another subject with force level changes of 0.2N. The ability of force regulation by the prosthetic finger was compared to that of the intact finger. Results showed that the overall root-mean-squares (RMS) error of the prosthetic finger was low, although significantly higher than the intact finger, 75% higher in subject 1 and 57% in subject 2. However, the correlation coefficient between the forces of prosthetic finger and intact finger was high, 75% for subject 1 and 84% for subject 2, respectively. This preliminary study is encouraging, illustrating the feasibility of accurate and stable control of different levels of fingertip forces by the prosthetic finger, which is comparable to that of the intact finger. This capability may allow the prosthetic hand for fine manipulation tasks, such as grasping brittle objects, or response to object slip during grasp.Clinical Relevance-This work attempts to restore the ability of a prosthetic hand for precision fingertip force control that may enrich the functionality for users in activities of daily living.