Advancing Human Fall Detection By Simulating Falls With A Ballistics Gelatin Torso
Grant Kuczler · Furman University Scholar Exchange (Furman University) · 2016
One in three senior citizens falls annually (National Center for Injury Prevention, 2015). These falls often go unreported and are a serious health and safety concern for the elderly (Independence, 2012). Companies, such as Life Alert, attempt to assist elderly people who fall but cannot help those who are unconscious or do not have their help button on them. A sensor-based program was developed to better assist fall victims. Using accelerometers, data acquisition software and a program to call for help, this system measures the acceleration in vibrations caused by movement around the sensors. The goal of the program is to distinguish among the vibrations made when someone is walking, bouncing a ball, has dropped something, or has fallen and needs assistance. This feature of the program, determining whether vibration pattern is a fall, is still being developed. Young, healthy volunteers have been used in past studies, but yielded dissimilar results to data collected from actual fall victims (Klenk et al., 2010). Beef gelatin, when mixed in the correct proportions, can accurately model the density of the human body. Jerk and acceleration are the two metrics used to compare this test’s results to real-world falls. The gelatin model generates similar acceleration signals, but very different jerk and variance of acceleration values. This project shows that gelatin cannot correctly model the human torso when only the density is adjusted. However, if the gelatin model is modified, by adding more structure, it could serve as one of the most accurate fall simulants.