A Formal Method of Measuring Interactivity in HRI

Abstract

In human-robot interaction (HRI), interactivity is a critical issue in terms of how they understand each other and what makes them share grounds to communicate. This paper proposes measures to evaluate interaction process using information theoretic approach. The measures are designed to quantify the level of shared ground among agents and their environment. With these measures, an experiment is carried out in which human subjects were asked to interact with a virtual robot to assign tasks with voice interface. The results of the experiments show that the measures are useful in measuring the interactivity of HRI.