Gaze-based hints during child-robot gameplay

This paper presents a study that examines whether gaze hints provided by a robot tutor influences the behavior of children in a card matching game. In this regard, we conducted a within-subjects experiment, in which children played a card game “Memory” in the presence of a robot tutor in two sessions. In one session (Help), the robot tutor gives gaze hints to help children find matching cards and, in the other session (No_Help), the robot tutor does not provide help to the children. Gaze hints involved looking toward the correct matching cart. We analyzed the child-robot interaction regarding execution performance, gaze behavior, and level of engagement. Children performance was measured using the number of trials and overall time used to complete the game. We found that children used significantly fewer trials in the Help condition than in the No_Help condition. In addition, there were more instances of mutual gaze in the Help condition than in the No_Help condition. These findings suggest that designing a robot with dynamic gaze increases the effectiveness of the robot tutor as a helping agent.