Challenges for Social Embodiment

Abstract

Current research in the area of social signal processing focuses on offline analysis of previously recorded human social cues. Approaches to exploit social signal processing techniques in naturalistic environments where agents socially interact with humans are rare and typically focus on isolated aspects, such as the creation of appropriate head nods or gaze behaviors. This position paper aims to identify challenges and research objectives for the area of social signal processing in order to encourage applications with more advanced forms of social embodiment in interactive settings.