Design and Development of Smart-Jacket for Posture Detection

Abstract

Complex human activity is hard to classify with observational logic. The advent of fabric sensors with discernible and steady response have opened a new avenue for classifying physical activity of humans. Our goal is to construct a smart jacket for human activity/posture classification and to apply machine learning models on the fabric sensor reading to predict physical events. The core concept is in placing stretch sensors, pressure sensors and accelerometer at strategic location to collect the responses. The sensor's response is studied toidentify linearity and repeatability, using which, reliability of data is determined. Further, appropriate Machine learning algorithms can be employed to classify different set of activities. It also important to follow a proper procedure to record data from fabric sensors which create a voltage fluctuation when stretched. We propose a systematic way of design, development, testing and integration of fabric sensors for reliable data collection in this paper.