Towards Partners Profiling in Human Robot Interaction Contexts

Individuality is one of the most important qualities of humans. Social robots should be able to model the individuality of the human partners and to modify their behaviours accordingly.This paper proposes a profiling system for social robots to be able to learn the individuality of human partners in social contexts. Profiles are expressed in terms of of identities and preferences bound together. In particular, people’s identity is captured by the use of facial features, while preferences are extracted from the discussion between the partners. Both are bound using an Hebb network. Experiments show the feasibility and the performances of the approach presented.