Manifestation of various adaptive motions by learning independent of environmental or oscillator models
| Project/Area Number |
18H01399
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Review Section |
Basic Section 20020:Robotics and intelligent system-related
|
|---|
| Research Institution | Tohoku University |
|---|
Principal Investigator |
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
大脇 大 東北大学, 工学研究科, 准教授 (40551908)
|
|---|
| Project Period (FY) |
2018-04-01 – 2021-03-31
|
| Project Status |
Completed (Fiscal Year 2021)
|
| Budget Amount *help |
¥16,380,000 (Direct Cost: ¥12,600,000、Indirect Cost: ¥3,780,000)
Fiscal Year 2020: ¥2,600,000 (Direct Cost: ¥2,000,000、Indirect Cost: ¥600,000)
Fiscal Year 2019: ¥4,550,000 (Direct Cost: ¥3,500,000、Indirect Cost: ¥1,050,000)
Fiscal Year 2018: ¥9,230,000 (Direct Cost: ¥7,100,000、Indirect Cost: ¥2,130,000)
|
|---|
| Keywords | 深層学習 / 運動学習 / 運動シナジー / バランス制御 / 筋電位 / 環境適応 / 歩行 / 姿勢制御 / 運動計測 / 運動適応 / 歩行制御 |
|---|
| Outline of Final Research Achievements |
To elucidate the mechanism in environmental model-independent adaptive learning, the mechanism of motion pattern generation was investigated in detail using model-free deep reinforcement learning. Although mathematical optimization calculations require a mathematical model of the environment and the body a priori, none have been treated so far as motor learning in an unknown physical environment. We investigated for potential computational guidelines in multi-joint gait. We examined the level of emergence of motor synergy in walking movements that changed as learning progressed and found that the synergy emergence was highly correlated with the performance per energy. It implies that motor synergy is employed as a necessary condition for improving performance per energy.
|
| Academic Significance and Societal Importance of the Research Achievements |
深層強化学習による運動学習タスクにおいて運動シナジーの発現プロセスが起きており、それがエネルギー当たりのパフォーマンスと高い相関を示したことは、何故人間や生物が運動シナジーを活用しているのかという問いの答えにつながるため科学的な意義が高いと考えられる。工学的な応用としては現在の深層学習は膨大な計算コストを要するが、効率的な運動学習における潜在的な方策として運動シナジーを用いることができたら大幅な計算の効率化につなげることができるため、本研究は新しい深層強化運動学習アルゴリズムに向けて示唆に富む情報となることが期待される。
|
Report
(4 results)
Research Products
(50 results)
