Theoretical Studies on Perceptual Computing and its Application to Human-Computer Interaction
The aim of this research is to make further advances in human perception-related information processing technologies as well as to develop highly user-friendly human-computer interaction systems which apply these technologies.
With regard to elemental voice and image processing technologies, methods for accommodating the diverse phenomena are being studied especially focused in conversational situation. In conversation, for example, utterance timing, prosody, voice property, facial expression, signs, etc., undergo a variety of changes depending upon the situation, and these changes convey important nuances. The mechanisms for understanding and expressing this information are being studied using advanced signal processing theory and machine learning theory. Also, research is being conducted to identify methodologies which will allow systems to not just simply exchange information but to also create a “pleasant conversational situation.”
One application being considered is a computer interface system for use primarily by the elderly. It is essential that the system is transparent in order to be user-friendly, that means the policy of the interface design should be simple and easy to understand by the elderly. Basic knowledge about the relation between the design policy of the interface and the user-friendliness is being accumulated. Research into how voice and image processing technologies can be used to exploit these findings is also underway. Two actual applications currently undergoing verification experiments in nursing care centers and other facilities; one is a system that allows the elderly to enjoy communicating with their families via Internet-based content and the other is a robot which plays a game with group of elderly and stimulates communication of them.