Experimental extraction of tactile sensor specifications for a minimal anthropomorphic robot hand

Abstract

In this work a new approach is taken towards determining the quantified contribution of tactile acuity to human manual dexterity, and the implications of this approach when applied to the development of artificial fingertip touch sensors for humanoid robots or for prosthetic hands. The interdependence between several dimensions of both tactile acuity and dexterity is investigated. An experimental study was performed on a carefully chosen sample of 30 human subjects, with data acquisition taking place over a total period of 35 hours of testing in a controlled environment. The data were analyzed to extract minimum levels of tactile acuity that would result in manual dexterity performance at 80% of normal. These extracted levels are interpreted to represent minimum specifications for the design of an artificial tactile sensor that would endow a robot hand with acceptable dexterity, and are used in a case study to drive the conceptual design process for a new tactile sensor based on quantum tunneling composite material.