Development of an Instrumented Cane for Real Time Balance Estimation ; リアルタイムバランス評価のための計装杖の開発

九州工業大学 ; 博士（工学） ; 1. Introduction||2. Previous Studies||3. Inertial Sensor-Based Walking Cane Motion Estimation||4. Walking Cane mounted Millimeter Wave Radar Characterization||5. Millimeter Wave Radar-Based Base of Support Estimation||6. Millimeter Wave Radar-Based Center of Gravity Estimation||7. Conclusions and Future Work ; Falls are a global health issue that particularly affect to the elderly. It is estimated that about 30% of people aged over 65 years will fall at least once per year worldwide. Moreover, falls in the elderly often result in injuries that affect to their quality of life and in some cases, it leads to deaths. Several studies have worked on the development of fall detection systems but only few have tried to develop systems that try to prevent falls by estimating the fall risk in real time. The purpose of this research is to develop an instrumented cane for real time balance estimation. The balance ability is one of the main indicators for the risk of fall, therefore our proposed system could be used as an estimator of the fall risk in real time. This research focuses on cane users because canes are the most common type of mobility aid for the elderly and yet few studies have focused on the balance analysis of cane users. Based on previous studies, the proposed system estimates balance by analyzing the position of the Center of Gravity （CoG） with respect to the Base of Support （BoS）. For this purpose, this research proposes a balance estimation system that consists of: 1） feet position estimation using a millimeter wave radar to obtain the BoS; 2） cane motion estimation using an inertial sensor to correct the radar’s data; and 3） CoG position estimation using the millimeter wave radar’s data. According to experimental results, the proposed system was able to estimate the BoS with a similar error compared to previous shoe type devices. Furthermore, the proposed cane motion estimation system had a higher phase estimation accuracy than previous methods using a lightweight Convolutional Neural ...