Fall Detection for Elderly

Abstract

In daily life, falling is one of the major reasons to cause personal injury, especially for the elderly. If there is no any action to take care of the person who fell, it may become seriously physical damage and fatal. Therefore, it is necessary to develop a system to monitor walking status of people in our daily life. In the project, the normal walking gait and motionless (e.g. standing) is investigated. Based on the understanding of walking gait, a fall detection algorithm is then developed to determine whether fall accident is occurred. To collect the data of the motion, a 3-axis-accelerometer sensor is used, which can measure the accelerations due to gravity in 3 axis (X, Y, Z). An Android application is developed to make the algorithm able to apply in daily life better. To analyze collected data, some method are used. A moving window method is used to do calculation (e.g. mean) to analyze the walking gait. To improve the accuracy to compute period, a low pass filter is used to smooth data. Some advanced calculation is developed to get specific data, such as mean of minimum peak. In the algorithm, normal status and falling can be determined.