1A1-M01 仮想物理環境におけるバッタモデルの跳躍行動の獲得(進化・学習とロボティクス)

Abstract

This paper proposes how to acquire successive jumping on artificial creature like a locust under the physical virtual environment. A concept of "behavior simple (BS) and behavior composed (BC)" is applied to the model. In order to realize successive jumping, its behavior is divided into two BSs, taking off and recovery motion. After two BSs are trained by use of the evolutionary artificial neural network (EANN), successive jumping can be acquired as BC.