Predictability in Human-Robot Interactions for Autistic Children

Abstract

A commonly used argument for using robots in interventions for autistic children is that robots can be very predictable. Even though robot behaviour can be designed to be perceived as predictable, a degree of perceived unpredictability is unavoidable and may sometimes be desirable to some extent. To balance the robot's predictability for autistic children, we will need to gain a better understanding of what factors influence the perceived (un)predictability of the robot, how those factors can be taken into account through the design of the interaction, and how they influence the autistic child-robot interaction. In our work, we look at a specific type of predictability and define it as “the ability to quickly and accurately predict the robot's future actions”. Initial results show that seeing the cause of a robot's responsive actions influences to what extent it is perceived as being unpredictable and its competence. In future work, we will investigate the effects of the variability of the robot's behaviour on the perceived predictability of a robot for both typically developing and autistic individuals.