In Japan's rapidly aging society, the number of elderly people living alone increases every year. Theses elderly people require more and more to maintain as independent a life as possible in their own homes. It is necessary to make living spaces that assist in providing safe and comfortable lives. "Biofication of Living Spaces" is proposed with the concept of creating save and pleasant living environments. It implies learning from biological systems, and applying to living spaces features such as high adaptability and excellent tolerance to environmental changes. As a first step towards realizing "Biofied Spaces", a system for acquisition and storing information must be developed. This system is similar to the five human senses. The information acquired includes environmental information such as temperature, human behavior, psychological state and location of furniture. This study addresses human behavior as it is the most important factor in design of a living space. In the present study, pyroelectric infrared sensors were chosen for human behavior recognition. The pyroelectric infrared sensor is advantageous in that it has no limitation on the number of sensors put in a single space because sensors do not interfere with each other. Wavelet analysis was applied to the output time histories of the pyroelectric infrared sensors. The system successfully classified walking patterns with 99.5% accuracy of walking direction (from right or left) and 85.7% accuracy of distance for 440 patterns pre-learned and an accuracy of over 80% accuracy of walking direction for 720 non-learned patterns.

Date Published: 8 April 2008
PDF: 10 pages
Proc. SPIE 6932, Sensors and Smart Structures Technologies for Civil, Mechanical, and Aerospace Systems 2008, 69320F (8 April 2008); doi: 10.1117/12.775707

Published in SPIE Proceedings Vol. 6932:
Sensors and Smart Structures Technologies for Civil, Mechanical, and Aerospace Systems 2008
Masayoshi Tomizuka, Editor(s)