2010 Fiscal Year Final Research Report
A Study on Robotic Attention for Action Detection based on Scanpath Theory
Project/Area Number |
20500182
|
Research Category |
Grant-in-Aid for Scientific Research (C)
|
Allocation Type | Single-year Grants |
Section | 一般 |
Research Field |
Perception information processing/Intelligent robotics
|
Research Institution | Tohoku Institute of Technology |
Principal Investigator |
FUJITA Toyomi Tohoku Institute of Technology, 工学部・知能エレクトロニクス学科, 准教授 (90293141)
|
Project Period (FY) |
2008 – 2010
|
Keywords | ロボット注視機能 / Scanpath / 関心領域 |
Research Abstract |
This study investigated human visual scanpath features in a scene of robot hand movement. Human regions-of-interest (hROIs) were measured in psychophysical experiments and compared using a positional similarity index on the basis of scanpath theory. Results showed consistent hROI loci due to dominant top-down active looking in such a scene. This study also presented a method for detecting hROIs which used both bottom-up and top-down image processing algorithms (IPAs). Algorithmic regions-of-interest (aROIs) generated by this method were compared to the hROIs. The results showed that the IPAs have a high ability to predict hROIs and to detect actions of a partner robot.
|
Research Products
(8 results)