| Project/Area Number |
16300058
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Perception information processing/Intelligent robotics
|
|---|
| Research Institution | Nara Institute of Science and Technology |
|---|
Principal Investigator |
MATSUMOTO Yoshio NAIST, School of Information Science, Assoc. Prof., 情報科学研究科, 助教授 (00314534)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
OGASAWARA Tsukasa Nara Institute of Science and Technology, School of Information Science, Professor, 情報科学研究科, 教授 (30304158)
UEDA Jun Nara Institute of Science and Technology, School of Information Science, Assistant. Prof., 情報科学研究科, 助手 (30346308)
UEDA Etsuko Nara Institute of Science and Technology, School of Information Science, Assoc. Prof., 情報科学研究科, 助手 (90379529)
HARADA Kensuke Nara Institute of Science and Technology, Intelligent System Research Institute, Researcher, 知能システム部, 研究員 (50294533)
TAKEMURA Kentaro Nara National College of Tech., Dept of EE, Assistant Prof., 電気工学科, 助手 (30435440)
|
|---|
| Project Period (FY) |
2004 – 2006
|
| Project Status |
Completed (Fiscal Year 2006)
|
| Budget Amount *help |
¥9,400,000 (Direct Cost: ¥9,400,000)
Fiscal Year 2006: ¥2,200,000 (Direct Cost: ¥2,200,000)
Fiscal Year 2005: ¥2,700,000 (Direct Cost: ¥2,700,000)
Fiscal Year 2004: ¥4,500,000 (Direct Cost: ¥4,500,000)
|
|---|
| Keywords | Humanoid Robot / Face and Gaze Measurement / Attention / Image Processing / State Estimation / Communication / Assistance / Video Lecture / 注視点 / 計測 / 画像 / 視線 / 推定 |
|---|
| Research Abstract |
1. Development of Measurement System for Multiple People
First, a method for measuring facial information using a single camera was developed. It employs the same type of 3D facial model as the one used in the conventional stereo camera based system. The optical flow fields corresponding to six DOF motions are generated using the 3D facial model. By assuming that an optical flow field due to arbitrary small 3D motion of the face can be represented by a linear combination of the six optical flow fields, the quantitative 3D motion of the face can be accurately estimated. Next, a method to estimate the focus of attention of multiple people was developed using a motion capture system and a stereo camera system. The focus of attention is represented as an ellipsoid whose position is defined as the intersection of the multiple gaze vectors and whose size corresponds to the concentration of the attention. Multiple attentions can also be estimated.
2. Communication Assistance based on Estimated
… More
Focus of Attention
The uses playing "word chain game" in a meeting-like environment is measured, and the focus of attention was estimated. The estimated attention was shown to be tracking the uttering person correctly. Based on the measurement method, automatic camera-work system is also constructed. Next, a method to estimate a user's behavior in video lecture was developed. By using gaze information, an object in attention was estimated, and by switching the bayse model based on the attention, a recognition rate of 85% was achieved. The estimated attention information and behavior information can be sent to the remote lecturer. Finally, a tele-communication using a robot as a media was investigated. The measured facial information (such as motion, gaze, and expression) was projected on a face robot, and it was shown that the system had a higher impact compared with video-chat and avatar chat systems. A tele-communication experiment using a humanoid HRP-2 was also conducted between Tokyo and Nara. In this experiment, it was shown that the remote user was able to obtain a wide field of view using a camera mounted on the head of the humanoid with the head direction controlled by his head motion. The other user sitting in front of the humanoid was able to understand the attention of the remote user via the motion of the robot. Less
|
