A Multimodal Dataset for Robot Learning to Imitate Social Human-Human Interaction

Humans tend to use various nonverbal signals to communicate their messages to their interaction partners. Previous studies utilised this channel as an essential clue to develop automatic approaches for understanding, modelling and synthesizing individual behaviours in human-human interaction and human-robot interaction settings. On the other hand, in small-group interactions, an essential aspect of communication is the dynamic exchange of social signals among interlocutors. This paper introduces LISI-HHI - Learning to Imitate Social Human-Human Interaction, a dataset of dyadic human inter- actions recorded in a wide range of communication scenarios. The dataset contains multiple modalities simultaneously captured by high-accuracy sensors, including motion capture, RGB-D cameras, eye trackers, and microphones. LISI-HHI is designed to be a benchmark for HRI and multimodal learning research for modelling intra- and interpersonal nonverbal signals in social interaction contexts and investigating how to transfer such models to social robots.