Site: Sony Computer Science Laboratory, Inc.
Takanawa Muse Bldg.
3-14-13, Higashigotanda Shinagawa-ku
Tokyo 141, Japan
Date Visited: May 25, 1995
Report Author: T. Skelly
The Sony Computer Science Laboratory (SCLC) was founded in February 1988 for the sole purpose of conducting research relating to computer science. The laboratory's objective is "to contribute extensively to social and industrial development through original research that looks ahead to the 21st century and has the potential for achieving breakthroughs in computer development."
While occupying a single floor and with a staff of only 16, SCSL achieves enthusiasm and productivity in what appears to be a relaxed atmosphere. Perhaps it is because results achieved by each member of the SCSL are evaluated based on technical papers, research software products, domestic and international conferences and professional associations, and other achievements. Compensation is based on achievement rather than on seniority. This has allowed SCSL to attract many young and highly talented researchers.
Each member of the SCSL team sets his or her own research goals within basic research themes. Mario Tokoro, director of the laboratory, has said, "Our work is unrestrained by commercial needs. With the policy of bringing out the best in individuals, we respect and foster each member's initiative and creative ability." Research currently underway at the SCSL is focused on distributed operating systems, computer networks, programming languages, human-computer interaction, artificial intelligence, complex systems, and other fundamental aspects of next-generation computers. The SCSL's research goal is "the development of technology enabling the creation of a safe, evolutionary, stable, cohabiting information society." Results achieved thus far include an object-oriented operating system for distributed processing, a mobile host protocol, a computational field model, and agent-based human-computer interaction. The first presentation that the JTEC team saw was by Ken Miyashita on "Interactive Generation of Graphical User Interfaces by Multiple Visual Examples." This work attempts to reduce the amount of coding necessary for graphical layout of relationships by allowing the system to interpret and recode itself, based on direct modifications to the system-generated visualization The second presentation was by Dr. Akikazu Takeuchi, who is doing research focused on facial expressions, voice tones, and gestures and how they express emotions and feelings. The SCSL staff believe that this research will help to introduce emotional factors into human-computer interactions. "In the future, a virtual human, which can interact emotionally with a real human through a display and a video camera, may be a reality."
Dr. Takeuchi showed videotapes of two examples of his system in trials. The first presented the face of his son (mapped onto a 3-D model) answering questions about new Sony products (see Fig. 4.5). This version of the system uses speech recognition and synthesis, syntactic and semantic analysis, plan recognition, response generation, and most interestingly, correspondence between conversational situations and facial displays. In short, the user asks questions of the boy's image on the computer screen. The system then interprets the question, plans an appropriate response, and animates the boy's image with both lip movement and emotional facial responses.
Dr. Takeuchi's second system involved two users playing a card game. The cards were displayed upon the screen, along with a woman's face. Through various facial gestures, the computer attempted to give hints and encouragement to one of the players. A video camera was used as input to detect the position of the players. Xerox PARC and the TRON project propose that computers should be embedded in nearly every object in our environment, perhaps even ourselves. A much less intrusive and practical approach to embedding knowledge in the real world was presented by Junichi Rekimoto. This researcher coupled a lightweight video display and camera to create a sort of looking glass, through which the user could view an environment with additional information overlaying the image when desired. By using simple image recognition technology, he was able to label real-world objects by means of hand-made bar-codes. While still very much a prototype, the clear concept was that for a computer to interact with the world, the world needs only supply an identification code, easily read by a hand-held reader. Privacy is maintained, as neither the world nor the user needs to broadcast any information other than a unique, machine readable symbol. It is entirely up to the software possessed by the user to gain any additional information, such as history, contents, or further instructions. The same label could trigger different information to different users, depending on their access to the proper identifying link between label and information. This clever idea, besides being inexpensive and multi-functional, also reduces the problem of system compatibility to that of creating standardized identifying labels and readers.
Sony's Computer Science Laboratory is a brave, albeit small, experiment. In recognizing the future role of computing in consumer products, Sony is risking going beyond Japanese corporate tradition in order to obtain what may possibly be exciting, breakthrough results.
Miyashita, K., S. Matsuoka, S. Takahashi, and A. Yonezawa. 1994. Interactive generation of graphical user interfaces by multiple visual examples. Proceedings of UIST.
Rekimoto, J. 1995. The world through the computer: Computer augmented interaction with real world environments. UIST '95 Proceedings.
Sony Corporation. N.d. Sony Computer Science Laboratory, Inc. Brochure.
Takeuchi, A., and K. Nagao. 1993. Communicative facial displays as a new conversational modality. Proceedings of INTERCHI '93.