Interactive Behavioral Design between Autonomous Behavioral Criteria Learning System and Human

In this study, we suggest a learning system for the autonomous creation of novel behavioral patterns through interaction with human beings. Two kinds of evolutionary calculation are used in the proposed system: genetic algorithms (GAs) with which the system learns behavioral patterns and genetic programming (GP) with which the system acquires behavioral criteria depending on the behavioral pattern learned with GAs. In addition, novel behaviors are designed with the learned behavioral criteria through interaction between the system and the user. Since the system generates new behaviors with new behavioral criteria instead of simply combining existing behaviors, various novel behaviors can be created. Computer experiments have been carried out to demonstrate the effectiveness of the proposed system.