Multimodal face-to-face interaction with a talking face: mutual attention and deixis
Résumé
Our long-term goal is to build an embodied conversational agent able to maintain realistic face-to-face communication with a human interlocutor. This conversational agent is embodied by a videorealistic talking head. While most researchers focus on discourse interpretation and generation, the main challenge here is to provide the interlocutor with implicit and explicit signs of mutual interest and attention as well as with an awareness of environmental conditions in which interaction takes place. A hybrid platform, with hardware and software, has been developed to test various interaction scenarios. As an application, the talking agent was used to interact with a user during a simple card game. The role of the agent was to act as a guide as well as to provide different levels of guidance (with or without mutual attention, and with or without endogenous eye saccades toward a correct or incorrect play). We provide here a comparative analysis of user performance across different levels of guidance and user perception of both level of guidance and level of help from the embodied conversational agent.
Origine : Fichiers produits par l'(les) auteur(s)
Loading...