Human behavior modeling for multimodal interaction with robot partner

If a robot partner can provide a multimodal interaction with a person, it may offer more smooth and natural human-robot communication. This paper proposes a method for modeling relationship between behavioral and emotional expression through human-robot interaction. First, we applied a neuro-fuzzy system to the classification of human posture. Next, we used the facial expression recognition algorithms based on Constrained Local Model and two-stage fuzzy reasoning model. Furthermore, we utilized spiking neural network for the relation modeling. In this paper, we show an experimental example to examine the proposed method.