|
Hironobu TakahashiTRC Active IntelligenceLaboratory |
| Affiliation: | Active Intelligence Laboratory, Theory and Novel Functions Department, Tsukuba Research Center |
| Birth date: | December 17, 1958 |
| Birthplace: | Miyagi |
| Background: | |
| 1982 | Graduated from the Faculty of Science, Kyoto University |
| 1984 | Completed the masterｹs course of biophysics, Graduate School of Science, Kyoto University |
| 1985 | Joined Sanyo Electric Co., Ltd. |
| 1993 | On loan to the Real World Computing Partnership |
I was born in Miyagi, a north-eastern prefecture on Honshu Island. Attracted by the Kansai Area in the western part of the main island including Osaka and Kyoto, I went down to the area and spent seven years in Kyoto, an ancient capital of Japan. When I went to university in Kyoto, I was ambitious to be a world-famous scientist and find a "better half." It is traditionally believed in Japan that a man from eastern Japan and a woman from western Japan make a very good couple. A few days after I entered the university, I abandoned my first ambition to be an internationally renowned scientist because I met many people far better than I. But I achieved my second goal by finding my wife. Half success and half failure.
I joined Sanyo in 1985 when the company was looking for staff for their newly-built research center in Tsukuba. Since then, I have been in this rural academic city for 11 years. In 1985 when I came here, I found the city of Tsukuba very attractive because it had a number of research institutes and a World Science Exposition was being held then.
Maybe because I was born and raised in a rural place, I feel comfortable in this open city and prefer it to the crowded urban areas like Kyoto where houses are jammed together with just a few inches' clearance. In Tsukuba, the streets are wide and there are many places where children can freely play. I'd like to believe that's the reason I had a fourth child last year.
I often spend my private time computer-hacking. In 1986, I was on loan to the Electrotechnical Laboratory for a computer vision study. My assignment to the institute coincided with the arrival of JUNET. I still remember that I was very impressed by the volunteer network as it covered the entire world.
During the mid 1980s, we had to use every trick we knew in order to run in Japanese environments the English applications and tools delivered by Usenet and Internet. So, I tried to localize Internet tools into Japanese as a volunteer and for pleasure.
Also, I created a utility software named "kakasi" as we needed to convert Japanese sentences with a mixture of kanji, katakana and hiragana into hiragana-only or alphabet-only sentences on network. It is not a sophisticated algorithm and would make Japanese processing experts mad. Yet, it is widely used for many purposes including the conversion of Japanese sentences into Roman alphabet or "romaji" for non-Japanese speakers and the creation of dictionaries with entries listed in the order of their pronunciation. You will find "kakasi" in the TRC environment as well.
When I was a university student, I majored in the physiology of the eye. But I spent most of my time in the computer room saying "this is an experiment." I decided to join the electric manufacturer as a computer vision researcher because I thought it would be possible to get involved both in the study of eye physiology and computer there.
When I started my career, researchers were trying to reproduce the ability of human sight by computer as part of artificial intelligence research as exemplified by "Vision," Marr's posthumous work. Image processing and the range finder had already been in wide use and were to be found in industrial applications. However, researchers were eagerly pursuing cognition methods and systems much closer to the human sense. That is why I have been continuously involved in research activities based on stereo vision. Once completed, I believe it will contribute to the development of artificial intelligence and intelligent robots in the research field and offer a number of industrial applications.
Stereo vision was almost blind ten years ago. Now I can use the resources of the RWCP as a member of the partnership, and this system can find and recognize a target on a desk in 10 seconds once an appropriate model is given. I also developed in cooperation with Dr. Suehiro, chief of the TRC Active Intelligence Laboratory, a system which can pursue in real-time an object by using its stereo sight.
At present, I am responsible for developing the cognitive capability of an agent network and am engaged in assembly work. It will be completed by the end of this year or soon after the beginning of next year.