MF133 Study on prediction of falling when using a walking aid
Yusuke Ouchi, Osamu Matsumoto, Seong‐Sik Yoon, Yoji Yamada, Yoshitaka Adachi · The Proceedings of the JSME Symposium on Welfare Engineering · 2007
In the study, we propose a risk prediction method which combines a motion prediction method based on a human physical model with a motion inference one that originates in the probability theory for a human falling prediction purpose when an elderly person is using a walking aid. We confirm the effectiveness of the proposed method by experiments using a sensor system attached to a subject. The target is limited to human falling in case he is using a walking aid. The method determines whether he is in the state of safety or danger by use of Hidden Markov Models with his position, velocity, and direction of C.O.G data introduced. Moreover, we developed a small, light-weight, and wireless sensor system attached to the subject, and confirmed the possibility of predicting a subject's fall through our overall proposal.