Measurement and Analysis of Human Behavior Using Wearable Sensors Made of Conductive Knit

Abstract

Recently, the number of senior citizens who live alone is increasing rapidly in Japan. In this paper, we propose a monitoring system that watches over everyday living activities of the elderlies by using textile wearable device for supporting their lives. Several activities have been monitored by installing some sensors, or by wearing some nondaily sensors on the body in a conventional method. However, the cost and the burden of these sensors are serious issues. We developed new conductive knit and clothes made of the knit. This clothes functioned as a sensor for measuring human trunk and limbs behavior by only wearing it, without any burden. First, we measure human behavior in living activities by wearing this cloth. Next, we recognized these activities from the results of measurement with Hierarchical Neural Network. In experiments, we succeeded in classifying eight living activities with an accuracy rate of about 90 percent.