A Computational Model of Imitation and Autonomous Behavior in Continuous Spaces

Learning is essential for an autonomous agent to adapt to an environment. One method that can be used is learning through trial and error. However, it is impractical because of the long learning time required when the agent learns in a complex environment. Therefore, some guidelines are necessary to expedite the learning process in the environment. Imitation can be used by agents as a guideline for learning. Sakato, Ozeki and Oka (2012) proposed a computational model of imitation and autonomous behavior. In the model, an agent can reduce its learning time through imitation. In this paper, we extend the model to continuous spaces, and add a function for selecting a target action for imitation from observed actions to the model. By these extension and adaptation, the model comes to adapt to more complex environment. Even in continuous spaces, the experimental results indicate that the model can adapt to an environment faster than a baseline model that learns only through trial and error.