Journals Proceedings

Abstract

Falling is a common accident and the most significant cause of injury for elderly person. This study investigates the methodology to identify falls from normal Activities of Daily Living (ADLs). In this study, a wireless sensor system (WSS), based on accelerometer and gyroscope, is placed at the centre of the chest to collect real-time fall data. The WSS contains a set of ADXL345 (3-axis digital accelerometer sensor), ITG3200 (3-axis digital gyroscope sensor), MCU LPC17680 (ARM 32-bit cortex M3), and Wi-Fi module RN13. Experiment protocols consisting of four types of falls such as forward fall, backward fall, and side way fall (left and right) along with normal gait involved 324 tests on 18 human subjects. The results from the experiment shows the system and algorithm could distinguish between falling and ADLs with an accuracy of 99.382%.