Period: August 22 - September 4, 1996(14 days)
Places (attended
conference):
1) ICPRš96 (International Conference on Pattern
Recognition)
2) SICS (Swedish Institute of Computer Science)
3) RISC-LINZ
(Research Institute for Symbolic Computation)
4) SNN (Foundation for Neural
Networks)
ICPR'96 (International Conference on Pattern Recognition), held biyearly and with this year the 13th, is a world-class conference on overall pattern recognition research, and this year it was held in Vienna. It was divided by subjects into 4 tracks including Computer Vision, Pattern Recognition and Signal Analysis, Applications and Robotic Systems, Parallel and Connectionist Systems, with 11 invited speeches, 63 oral presentations, and 452 poster presentations. On this trip, I also visited three research institutes: SICS (Swedish Institute of Computer Science), RISC-LINZ (Research Institute for Symbolic Computation), and SNN (Foundation for Neural Networks). I made presentations at those research institutes and the conference on a robot positioning method using motion images. I demonstrated real-time gesture recognition with video clips, and emphasized the usefulness of continuous DP used in both systems.
I was accompanied by Dr. Hiroshi Yamakawa of the TRC Active Intelligence Laboratory. He is an experienced traveler and was quite calm both during the take off and landing of our plane, so I was relaxed and enjoyed the trip in the sky.
We first visited SICS. Dr. Martin Nilsson spoke fluent Japanese, because he studied at a university in Japan. He is Japan-savvy in that he brought back a small tatami mat by plane and installed it in his apartment. He introduced a snake robot. Currently he is building one such robot with tens of joints, and conducting a simulation of reinforced learning taking into account the friction with the ground.
Then, we participated in ICPR'96 held in Vienna. My presentation was given on a poster. I was very nervous because it was my first time to give an official presentation in English. Dr. Yamakawa also gave a poster presentation in the same time slot as mine, but he was calm and took pictures of me doing my presentation. During the presentation, I was asked about omni-directional visual sensors and real-time gesture recognition, so I felt many were interested in this technology.
Then, we visited RISC-LINZ located in Linz, Austria. It had been a overseas subcontractor of RWCP up to last year, and we discussed the research status with Dr. Witold Jacak and other members. They introduced a technology to swap symbolic computation and neural network computation, and a robot arm with a neighboring sensor. Although RWCP's subcontract has terminated, they hope to promote human and technical transfers with Japan on multimedia. My presentation elicited a frank and tough question, relating to the necessity of visual guidance of robots, and questioned its superiority over the conventional internal sensor. To this question, I emphasized that a robot can estimate its position correctly even at power on and on unfavorable floors, if it can determine its position from images like a human can determine his or her position by looking around at the surroundings.
Finally, we visited SNN in Amsterdam. SNN is one of the RWC Distributed Laboratories. We discussed multiscale computer vision and autonomous navigation control using optical flow with Dr. Ben J. A. Krose, Dr. Bart ter Harr Romeny, and others.
During this trip, I got acquainted with researchers at ATR and Osaka University, known to be quite active in gesture recognition research. I enjoyed the active atmosphere of the international conference and discussed with researchers at the RWC Distributed Laboratory and Overseas Subcontractor, which have had a good influence on my research work.
