正確的跌倒偵測,可以使因跌倒而陷入無法自主行動的老年人,及時接受醫療照護,提升居家照護的品質。本研究提出一個能快速判定跌倒的偵測系統,使用Arduino開發板,搭配三軸加速度計放置於胸口,雙軸陀螺儀放置於左右手腕,透過Zigbee無線傳輸,對著偵測伺服器發送感應資料,由偵測伺服器整合資料後判別跌倒,實作出能準確分辨出是跌倒的後坐下跌倒、後躺下跌倒甚至是跳躍落地後滑倒與非跌倒的快速坐下、快速躺下、跳躍落地的跌倒偵測系統。本系統跌倒偵測的平均敏感度為90.666%,平均變異度為99.111%,非跌倒的判定時間約為0.2秒,而跌倒的判斷時間約為1.4秒。
Accurate fall detection can help the elderly adults that are injured after falls receive treatment in time and hence improve the quality of home care. This research presents a fall detection system that can catch fall events quickly. It combines Arduino development boards and a triaxial accelerometer placed in the chest and two biaxial gyroscopes worn in the wrist of each hand. The boards transfer data to a fall detection server through ZigBee. The fall detection server analyzes data, and then differentiates between fall events and non-fall events quickly. Experimental results of presented system show that the average sensitivity reaches 91% and the average specificity reaches 99%. Moreover, the time for determining non-fall is 0.2 seconds and 1.4 seconds for fall detection.