Towards socially-competent and culturally-adaptive artificial agents

Abstract The development of artificial agents for social interaction pushes to enrich robots with skills and knowledge about (local) norms. One possibility is distinguish the expressive functional orders during a human-robot interaction. overarching aim this work set framework make agent socially-competent beyond dyadic – in varying multi-party situations individual-based user personalization, thereby enlarging current conception “culturally-adaptive”. core idea provide capability handle different kinds interactional disruptions, associated recovery strategies, microsociology. result obtained by classifying investigating requirements robot’s architecture should satisfy exploit such knowledge. paper also highlights how level competence achieved focusing on just three dimensions: (i) capability, (ii) relational role, (iii) proximity, leaving aside further complexity full-fledged human-human interactions. Without going into technical aspects, End-to-end Data-driven Architectures Modular are discussed evaluate degree which they can new cultural Finally, list general proposed.