Human emotional understanding for empathetic companion robots

Abstract

Companion robots are becoming more common in home environments, as such a greater emphasis is required on analysis of human behaviour. An important aspect of human behaviour is emotion, both the ability to express and comprehend. While humans have developed excellent skills in inferring the emotional states of their counterparts via implicit cues such as facial expression and body language, this level of understanding is often neglected in Human Robot Interactions; furthermore, humans are able to empathetically respond to the emotions of others to create amore harmonious and person relationship. This paper is a preliminary proposal of a novel approach for facial emotional detection and appropriate empathetic responses, in conjunction with long term emotion mapping and prediction; the proposed system will be implemented on a social mobile robot, thus allowing a further level of behavioural comprehension to achieve a more human like encounter. The technique will be based on Fuzzy Cognitive Maps, using FACS Action Units as inputs, a high level facial descriptor layer and output of six emotions.