1P1-E01 感情表現動作の自律生成に向けてI : GAを用いたロボットの多様な動作生成

概要

論文の詳細を見る
This study develops a two-step genetic algorithm-based method to generate robot actions varied in terms of their impressions given to users. This study is a part of the project that aims to achieve autonomous generation of emotional actions for robots. This study aims at generating sampled actions that satisfy hardware constraints of the robot in concern and give varied impressions as well. The variation in the sampled action group was supported by the result of a survey. Two action groups were shown to subjects; one group consists of six actions selected by the proposed method, and the other one consists of six actions randomly selected from the actions that satisfy hardware constrains of robot. The subjects evaluated variety of each action group in five levels. The result showed that the action group selected by the proposed method was scored better evaluation points than the random group.
一般社団法人日本機械学会の論文