2004 Fiscal Year Final Research Report Summary
Construction of Intelligent Information Environment Based on Active Interaction
| Project/Area Number |
14550352
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
情報通信工学
|
|---|
| Research Institution | The University of Electro-Communications |
|---|
Principal Investigator |
KANEKO Masahide The University of Electro-Communications, Faculty of Electro-Communications, Professor, 大学院・電気通信学研究科, 教授 (90262039)
|
|---|
| Project Period (FY) |
2002 – 2004
|
| Keywords | Intelligent Information Environment / Active Interaction / "Kansei"-like Response / Mobile Robot / Localization of Sound Source / Human Interface / Integration of Audiovisual Information / Interactive Agent |
|---|
| Research Abstract |
In order to construct the intelligent information environment based on the concept of active interaction, the following studies were carried out. Possibility of a new style of user-robot interaction has been demonstrated and its usefulness has been proved through various experiments.
1.Active interaction based on ambiguous words indicating spatiotemporal position of object has been proposed. The method to interpret the words such as "a little more right", "last", and so on has been devised and its usefulness has been confirmed through computer simulation and experiments in a real environment. A new type of interaction reflecting the relative positions of user, robot and objects in the environment has been also proposed.
2.Various types of active interactions based on integration of audiovisual information have been studied intensively. Major topics include the followings : (a)Highly precise sound source localization using the integration of visual and sound information, (b)Forming share
… More
attention between robot and multiple users in conversational scenes, (c)Detection of speaker in alternative conversational scene based on the probabilistic integration of audiovisual information, (d)Interpretation of ambiguous indications in space using multimodal information and its application to the control of mobile robot, and (e)Tracking of moving humans by robot using motion and color features in images and estimation of sound source position.
3.A series of recognition of user's face action, robot's response to the recognized result, and user's response to robot's action has been organized to study "Kansei"-like response by robot. This experiment has revealed the importance of the bidirectional communication between user and robot.
4.Interaction agent using facial caricatures has been studied as one of the influential user-friendly interface in the intelligent information environment. Not only the synthesis of caricature from photographs, but also correspondence between words representing facial impression and caricature has been examined. Less
|
