10212 Recognition of Cyclic Gestures for A Partner Robot Based on Interactive Learning

Abstract

This paper proposes imitation learning for a partner robot by using human cyclic gestures. The mobile robot used as a partner robot must make decisions suitable to the facing environment, imitative learning does not explicitly need an evaluation function specific to each behavior. Therefore, the robot can directly learn behaviors through interaction with a human. In this paper, we propose imitative learning based on spiking neural network, self-organizing map, and steady-state genetic algorithm. Furthermore, we show experimental results of the partner robot based on imitation.