Study on Action Recognition and Autonomous Work by a Mobile Robot based on Attention Mechanism
| Project/Area Number |
15K00364
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Research Field |
Intelligent robotics
|
|---|
| Research Institution | Tohoku Institute of Technology |
|---|
Principal Investigator |
Fujita Toyomi 東北工業大学, 工学部, 教授 (90293141)
|
|---|
| Project Period (FY) |
2015-04-01 – 2019-03-31
|
| Project Status |
Completed (Fiscal Year 2018)
|
| Budget Amount *help |
¥4,940,000 (Direct Cost: ¥3,800,000、Indirect Cost: ¥1,140,000)
Fiscal Year 2017: ¥520,000 (Direct Cost: ¥400,000、Indirect Cost: ¥120,000)
Fiscal Year 2016: ¥520,000 (Direct Cost: ¥400,000、Indirect Cost: ¥120,000)
Fiscal Year 2015: ¥3,900,000 (Direct Cost: ¥3,000,000、Indirect Cost: ¥900,000)
|
|---|
| Keywords | 注視特性 / ロボット注視機能 / 自律運搬作業 / ロボット視覚 / Scanpath / 注視機能 / 関心領域 / 自律作業 / 他者動作認識 / 眼球運動 / 視線計測 |
|---|
| Outline of Final Research Achievements |
An attention mechanism is useful for obtaining essential information effectively in a cooperative work by multiple robots. Thus, an experiment was conducted to investigate characteristics of human attention when observing movies of robot action. The result suggested that we attend between objects and robot repeatedly to recognize the action, then attend the target object to anticipate the action.
Next, a method for generating robotic attention was considered by applying image processing techniques to detect 3-D position of a partner robot as a part of action recognition. Experimental result confirmed that the proposed method can detect valid position.
Also, two methods were presented for detecting attentive positions on a box-shaped object to grip and lift up in a transportation task based on corner detection and plane detection. These methods were applied to a tracked mobile robot with multiple manipulation arms. The experimental results showed that these are valid for autonomous work.
|
| Academic Significance and Societal Importance of the Research Achievements |
本研究でロボット動作観察時の注視特性の知見が得られた．今後の研究の発展でロボットによる効率的な情報取得が可能になってくることが期待できる．
さらに，画像処理に基づく注視を生成し，それにより他者ロボットの３次元位置の検出を行うことができた．また箱状対象物の自律把持持ち上げを実現することができた．今回考案した手法を応用していけば，協調相手のロボットの動作理解や，危険な現場での自律的な作業の実現が期待でき，将来の幅広いロボットの活躍が可能となってくる．
|
Report
(5 results)
Research Products
(16 results)
