623 手動による倒立棒の安定化制御における個人差・習熟度の獲得とカオス・エントロピ解析 : ニューロ制御器とファジィ制御器の同定器としての比較(ヒューマンモデリング,OS-7 ヒューマンダイナミクス,総合テーマ「伝統を,未来へ!」)

元データ 2010-09-14

概要

This paper tried to identify the individual difference and the proficiency of human operator's behavior from time series data by using fuzzy controller and neural controller acquiring individual skill and proficiency of human operator. The operators in the experiment are skilled to some extent in stabilizing the inverted pendulum by training, and the data of ten trials per person were successively taken for an analysis, where the waveforms of pendulum angle and cart displacement were recorded. The maximum Lyapunov exponents were estimated from experimental time series data against embedding dimensions. The identified fuzzy controller and neural controller from time series data of each trial of each operator showed well the human-generated decision-making characteristics with the chaos and the large amount of disorder. The estimated degree of freedom of motion increases and the estimated amount of disorder decreases with an increase of proficiency both in the simulations with identified fuzzy or neural controllers and the experiment. It also showed that the fuzzy controller is good for identification of initial stage and the neural contoroller is good for that of proficient stage of skill up process.

著者

川副 嘉彦 埼玉工業大学工学部
石川 恵介 埼玉工業大学大学院工学研究科
川副 嘉彦 埼玉工大
石川 恵介 埼玉工大
太田 智洋 坂本工業

関連論文

▼もっと見る