|
Fumio TogawaNovel Functions Sharp Laboratory |
| Affiliation: | Visual Media Research Institute, Engineering Division, Sharp Corp. |
| Date of birth: | July 31, 1953 |
| Birthplace: | Shiga |
| Background: | |
| 1976 | Graduated from the Electronic Engineering Department, Faculty of Engineering, Kanazawa University |
| 1976 | Joined Sharp Corp. Assigned to the Nara Plant |
| 1989-1991 | Engaged in joint development project of neuro-computer in Oregon |
| 1992 | Assigned to Makuhari Research Institute |
| Hobby: | Tennis, jogging |
After graduating from Kanazawa University, I joined Sharp Corporation in Nara, one of the old capitals of Japan, and was assigned to the development of information-related systems such as a speech recognition system. I can still remember with excitement the moment when the Japanese phrase meaning "He is ..." in Chinese characters appeared on the computer display immediately after I pronounced the phrase. Then, I developed a system which can understand hand-written characters. This technology is now employed in commercial products such as the "Shoin" word processor and the Zaurus PDA.
In 1987, my mother died and I wished I had been able to store electronically all of her memories. Around that time, I found myself interested in information processing of the human brain. I carefully read Professor Hinton's papers that Dr. Kai Fu Lee sent to me. Also, I read in turn original PDP texts with my colleagues. In 1989, I was allowed to participate in a neuro computer development effort with a venture business in Oregon, USA. During the two-year project, I was mainly engaged in the development of software and had opportunities to learn about many areas including SIMD architecture.
In 1992, I was assigned to the development of multimedia at a newly built institute in Makuhari, Chiba. That assignment led to my present mission at Novel Functions Sharp Laboratory in the RWC Program the following year.
My research theme, "Multimodal Human Interface with a Secretary Agent," is to develop a comprehensive interface which can be intuitively used, appealing to the five senses and very friendly to the user. In other words, it is a "multimodal human interface which allows communication with computers on a familiar daily conversational basis."
As an approach to this aim, I collect and build up a database of various motions and speeches in our daily conversations, analyze and model conversation patterns and verify the model using an experimental system. As a result of the repeated use of this three-step approach, I have created several experimental systems, including one which can play `scissors-paper-stone' at a human partner's pace and another which can recognize the human partner's intention through his/her tone of voice, eye contact and head motion and spontaneously respond to him/her. Using these systems, I have carried out a number of verification tests.
My next theme is how to optimally integrate different types of information along with the passage of real time by using voice, images, speech and virtual reality technology. So, I am now trying to solve, using a multimodal approach, the problems of "rhythm" and "emotion" which are beyond the capability of existing interfaces.
As the history of computer developments shows, we have adopted our own motions of pushing with a finger and writing on paper into interfaces such as the button, keyboard, mouse and pen. In our present research, other daily behavior during conversation will be the base for a computer interface.
My present goal is to establish a fundamental technology for an "international standard multimodal (agent) interface" which allows everyone to take advantage of information networks.