Fuzzy and Neural Computing for Communication of a Partner Robot
Naoyuki Kubota, Daisuke Hisajima, Fumio Kojima and Toshio Fukuda
We show how to apply fuzzy inference to design of communication between a robot and a human. At first, we discuss communication based on the concept of a perceiving-acting cycle, adopted from ecological psychology. Then, we apply fuzzy inference and modular neural networks to behavior learning of a partner robot. The fuzzy inference system is used in gaining temporal behavior knowledge for the partner robot, while modular neural networks are used to develop learning action rules through interaction with the human. Finally, we discuss the communication of the partner robot with the human in navigation experiments.