An Emotional Robot

The method to construct an emotional robot based on regulation of emotional responses with an emotion state embedded reinforcement learning system is proposed in this paper. Besides environmental states, the emotional robot has emotional states which are generated by stimulus received from sensor images. If the learning coefficient of emotions in Amygdala model is changed, generated emotional states in the robot are also different, even if the robot sees same sensor images. As a result, using the method, we can make kinds of robots with any emotions, having same structures. Through computer simulations, applying the proposed method to construct emotional robots, it is said the robots solve the path-finding problem including a variety of distinctive solutions. We find that each robot is able to have each individual solution depending on kinds of its emotions.