The role that the internal model of the other play in learning the cooperative behavior

摘要

In this first study, the simple cooperative behavior is to move a stick object from a start point to a goal point by two robots. The behavior of one robot is controlled by deterministic simple rule. Human subject controls the behavior of another robot. The experiment is done under two conditions given to the subject. This result suggests that the modeling of another robot is essential for a successful cooperative behavior. Next, a module to predict the motion of another agent is added to the network, which has a potential to learn an internal model of the other agent. The robot acquired cooperative behavior.