• personality model; social robotics; expert systems

Designing the personality of robots is an emerging research trend in the field of Social Robotics. This topic requires the coupling of several disciplines to address the challenge, such as robotics, computer science, engineering, psychology, and neuroscience. Nonetheless, due to several conceptual issues and concrete implementation difficulties, there are not coherent guidelines to follow at present.
The personality model developed in this work is inspired by the social-cognitive theory of Mischel and Shoda. In contrast to the large majority of published works that employ findings from trait theories of personality, this theory focuses on the social and affective dimensions.
The model is conceived as a rule-based expert system implemented with the CLIPS production system. The result is a net of specific cognitive-affective units that generate the personality of the social robot.
The overall model is integrated into SEAI (Social Emotional Artificial Intelligence), a framework designed for controlling social robots. The model also includes an internal emotional variable that follows the Russell circumplex model of affects. This internal representation allows the modulation of the robot personality by accounting for the affective context of interaction.
Finally, the model has been tested in a social scenario with the humanoid robot FACE (Facial Automaton for Conveying Emotions), previously developed at the Centro di Ricerca “E. Piaggio” of the University of Pisa. Thanks to the application of the presented model, FACE was able to execute, in specific social contexts, the appropriate behaviours depending on the intended personality.