We All Make Mistakes: Terminal, Non-critical, Recoverable, and Favorable Interaction Failures Between People and a Social Robot
In this paper, we present an in-depth illustration of interaction failures relatively unexplored in the field of human-robot interaction (HRI). Our qualitative analysis of interactions between a social robot and 12 participants sheds light on different types of erroneous interactions initiated by human and robot actors and their outcomes. Our findings show that a small portion of observed failures had fatal impacts on interactions. In most cases, they had little negative effects on interactions or even led to favorable outcomes, causing laughter and giggling from participants, for example. Overall, our study calls for further examination of the roles of failures and contextual factors that influence the consequences of failures in HRI.