As the society ages, the number of falls and fractures suffered by the elderly is increasing significantly in numbers. However, studies with reliable statistics and analysis on falls of this specific population were scarce. Fractures due to falls of the elderly are potentially of critical severity, and, therefore, it is important to detect such incidents with accuracy to prevent fractures. This necessitates an effective system to detect falls. For this reason, we induced simulated falls that resemble actual falls as much as possible by using a fall-inducing apparatus, and observed the movement of the body during the falls. The movement of the body was sensed using 3-axes acceleration sensors and bluetooth modules, which would not obstruct the movement as wired sensors or movement analysis systems would do. Using the acceleration data detected by the sensors, a fall detection algorithm was developed to detect a fall and, if any, its direction. Unlike existing studies that used sum-vectors and inclination sensors to detect the direction of falls, which took too much time, the system developed in this study could detect the direction of the fall by comparing only the acceleration data without requiring any other equations, resulting in faster response times.
Implementation and Application of Fall Detection on Mobile Perception Robot
