Audio-Based Analysis of Child-Robot Interactions in the Wild

We describe 2 in-the-wild child-robot interaction English-language group experiments conducted in India, using the social robot Vector. We analyze the children’s engagement with the robot for the duration of each interaction by analyzing their vocal behavior using 3 approaches: the gendering of the robot by the children, a behavioral categorization system, and a qualitative breakdown of vocal behaviors observed. We highlight the potential of audio- and language-based behavioral analysis for human-robot interaction.