Giving robots a flexible persona: The five factor model of artificial personality in action

A computational framework for artificial personality in cognitive robots is introduced. While every robot has some form of personality, the framework reported here is flexible and enables the exploration of different behaviors on the same robotic platform. The framework described here maintains a probabilistic representation of an internal state that includes emotion, motivation, sensing, and previous action. The next action is computed by using a massive number of rules implemented using Bayes Rule. This flexible Bayesian representation of personality allows the robots personality to be designed by a personality generator algorithm. The authors present results in a real robot and compare the behavior of robots with differing personalities.