A simple model for human-robot emotional interaction

This paper describes a model of human-robot interaction. It computes the robot's simulated emotional state as a result of the emotional input received from the user and the robot's internal emotion dynamics. The latter is determined by the key elements of personality and attitude. The proposed model is a probabilistic finite state automaton, where the probabilistic transition function is called personality and is modified during the interaction according to the history of the interaction itself. The criterion which drives this personality change is called attitude. Such a model may be used in affective computing applications in order to design emotional agents and robots.