Multi-agent System Engineering for Emphatic Human-Robot Interaction

Abstract

Human-robot interactions have to take into account the natural multi-modal bidirectional communication model that is common among humans. The model does not rely just on speech and verbal exchange, but it shall include emotional exchange through different channels: face muscles, body posture, voice modulation, skin responses, odors, etc. While some aspects are feasible yet far from being adopted by daily robotic interaction with humans, the other ones can exploit current level of technology so as to be included in common, although complex, human-robot communication use cases. In order to cope in synergic but efficient and modular way with the various emphatic communication aspects, we propose to employ intelligent agents and multi-agent system. Such multi-agent system comprises a controller sub-system aboard the robot, which is coordinated by logical agents that can incorporate perceptive modules which generates state predicates, reason about them, plan, and deliver emotionally intelligent action while interacting with human beings, emulating as much as possible human empathy.