The Development of a Facial-Affect Recognition System for Application in Human-Robot Interaction Scenarios

This paper presents a non-contact unique automated affect recognition system that identifies human facial expressions and classifies them using Support Vector Regression (SVR) into affective states based on a pleasure-arousal two-dimensional model of affect. By utilizing a continuous two-dimensional model, rather than a traditional discrete categorical model for affect, the proposed system captures complex and ambiguous emotions that are prevalent in real-world scenarios. Our aim is to incorporate the proposed recognition system in robots engaged in human-robot interaction (HRI) scenarios. Namely, the system can be utilized by a robot to recognize, in real-time, spontaneous natural facial expressions of a variety of individuals in response to environmental and interactive stimuli. Preliminary experiments demonstrate the system’s ability to recognize affect from a number of individuals displaying different facial expressions.