Smart Detection Algorithm of Falling Motion

Abstract

Falling is a very dangerous and involuntary motion of human being, which can happen in everywhere, every moment. No matter whatever causes, either adventitious things or nervous system problems, the motion is physically harmful to people. Falling motion can cause serious damage to people especially to elderly and pregnant women. To minimize the risk of falling, we need to study the falling motion of human being. Therefore, this project is about using a computer program to distinguish human motion. The goal of this project is to find a smart detection algorithm of falling motion. In this project, a tri-axial accelerometer is used to record the motion of human. In order to increase the accuracy of this algorithm, almost 30 volunteers were involved in the experiment. All volunteers were told to do a set of motions, including walking, sitting (hardly), jumping and falling. The motions can be distinguished by their specific and regular patterns. The algorithm is not only based on the threshold but also pattern recognition. Before starting the analysis, the user should calibrate the program first in order to get more accurate results. This detection program was examined and integrated into a user friendly interface. Although it is not a real time detection algorithm, the results are reliable.