Thirteenth International Conference on The Constructed Environment

Abstract

In the U.S., fall is the leading cause of pedestrian injuries in the built environment. To reduce pedestrian falls, wearable motion sensors, such as inertial measurement unit (IMU), have been applied for continuous, and less-invasive fall risk monitoring during people’s daily walk in the built environment. However, solely monitoring kinematic movement may not be able to differentiate fall risks from gait irregularities of normal ambulatory activities such as direction and pace changes. Therefore, this study applies a wristband-type electrodermal activity (EDA) biosensor together with a belt-clip-type IMU to account for both people’s stress response and gait irregularity to fall risks. A mixture model-based algorithm was applied to detect data points of abnormally high stress and gait irregularity respectively. Then, the co-occurrence of high stress and gait irregularity at the same time was identified as representing a fall risk occurrence. For validation, 30 subjects were asked to walk over a pre-determined route along which different types of fall hazards were positioned such as cluttered and slippery surfaces and poor lighting. Meanwhile, their EDA and IMU signals were collected with the two wearable biosensors. The proposed technique was applied on the collected signals and the authors found that fall risks the subjects experienced on the hazards were accurately identified with the proposed technique. This finding can significantly contribute to reducing fall accidents in the built environment by enabling us to monitor individual pedestrians’ fall risks and proactively conduct interventions, such as providing fall hazard-free routes and fixing hazards, before actual falls happen.