Live generation of interactive non-verbal behaviours

Psychology, robotic and virtual agents communities commonly claim that to enable natural interaction to take place within dyad of agents, the dyad must be the siege of dynamical coupling: to give to their partners a feeling of natural interaction, interactants be human, robotic or virtual, must be able to make the dynamic of their behaviour emerge both from their own internal states and from their partner’s behaviours. However, most virtual agents engines for interactions model communication as a step by step phenomenon, where pre-scripted signals and corresponding feedbacks alternate. We propose here an agent architecture which generates non-verbal behaviours in live, influenced by both the internal state of the agent and the continuously incoming reaction of its partner. This architecture enables an agent facing either another agent or a human, to emphasise shared behaviours (called Snowball effect), to decrease un-shared behaviours as well as to align dynamically with its partner’s behaviour.