In recent years, there has been a great deal of efforts to deal with the notion of "emotion" from an engineering point of view. In service robotics systems, the man-machine interaction becomes also unavoidable and there arises a sharp need for increase of understanding human emotions. Most of the available results are concerned, however, with off-line type of emotion recognition, and an emotional action model of the agent itself in a free set manner. Therefore they may show very limited applications if the subject matter is how the robot helps the human.
In this thesis, the notion of "emotion" to service robot is applied. By proposed the neural network structure, it is described that how the robot would understand the user`s emotional condition and show its reaction, depending on the user`s emotional state as a service robot. The proposed model has a property that the robot can recognize the user`s emotional state and ingratiate the user with learning capability in real-time and in unstructured environments.
As the Service Robot is to be designed to behave in consideration of the user`s emotional condition, this robot will be able to help human mentally and emotionally.