Autonomous Behavior Generator for a Companion Robot "SELF" with which Humans Do Not Get Bored

Abstract

In this paper, we propose a robot behavior generation scheme that generates behaviors based on the senses of "curiosity" and "boredom," to create a companion robot named SELF with which humans do not get bored. The scheme was developed using a simple but unique simulation environment. A red ball and a blue ball were displayed on the computer, and the human subject moved the red ball. A behavior generator was designed based on natural inter human interaction patterns, utilizing conditional probability. The autonomous learning capability of SELF was realized by updating this conditional probability. Our studies reveal that interaction capability and animal-like speed are required for a companion robot. In addition, this paper shows how SELF can have a sense of curiosity and boredom as seen in animals, including humans, by using the developed scheme. The impressions of the robot behavior generated by the proposed scheme were examined using the simulation environment where the blue ball was controlled by the proposed scheme.