the University of Tokyo
| JAPANESE |
Graduate School of Information Science and Technology, The University of Tokyo
Access & Campus Map
Education & Research
Introduction of IST
Departments and Faculty
 
Computer Science
  Mathematical Informatics
  Information Physics & Computing
  Information & Communication Engineering
  Mechano-Informatics
  Creative Informatics
International Center for Information Science and Technology (ICIST)
International Cooperation
Admissions
Home > Professors
Professors
2010/09/03

Intelligent robots that act through language understanding

Lecturer Wataru Takano
(Department of Mechano-Informatics)

While human motion is structured by its internal dynamics, human society, culture, thought, and communication are institutionalized and structured by the symbol system called “language.”  In other words, we live on various structures such as body and language.  Intelligence to capitalize on these structures is necessary for robots and humans to coexist.  Lecturer Takano succeeded in equipping a robot with a functionality to acquire human motion as symbols by imitating body language and gestures.  Motion pattern data is learned based on statistical models, which are called motion symbols and a robot recognizes a variety of motion as symbols and generates its own motion by accumulating these symbols.  The next goal is to add language as a means of communication.  This is a unique approach of enabling a robot to interpret motion as language and is the first step towards an intelligent robot that behaves through language reasoning.

If an elderly person gestures “I want that,” a robot will intuitively understand the request “OK, he wants a mobile phone” and bring it to him as an example of information assistance.
The biggest challenge in realizing this vision is to identify and understand how real-world data and language are connected.  Even if the connection is unraveled, another challenge is to understand how human language reasoning works.

Lecturer Takano introduced his own model.  For example, the symbol “walk” can lead to associated words like “pedestrian,” “human,” and “road” using a model to associate words from accumulated motion symbols.  Then, a morphological analysis model from natural language research is used to generate sentences by grammatically rearranging them.  These two models are integrated to tackle difficult challenges.


Graduate School of Information Science and Technology
the University of Tokyo