ART: Adaptive and Real-time Fall Detection Using COTS Smart Watch

Fall detection is one of the most important issues for taking good care of the elder people. Previous solutions are mainly focusing on the fall detection accuracy to avoid unnecessary false alarm with wearable devices, which provide the detection ability after the fall events. In this study, we propose ART, an accurate and real-time fall detection algorithm using COTS smart watch, which could provide accurateter that, we extract the threshold of the acceleration and angle by using the method of SVM. Evaluation results show that, ART can detect fall effectively before the impact, and the accuracy of the forward fall, backward fall and the lateral fall are 100%,95.5% and 95.7% respectively. The average processing time is 276ms, which can provide relatively sufficient time for fall prevention.