Abstract
Toe clearance and MTC are important parameters related to fall-risk. MTC is minimum toe clearance during the swing phase. For daily usable walking assistance system, this study aims to estimating the toe clearance from inertial sensors, such as Inertial Measurement Units (IMU), attached to the thigh and shin instead of the feet. For this purpose, neural network system was introduced; however, differently from conventional data analysis, FFT analysis was conducted to IMU sensor output. This proposal is from considering the periodic characteristics of walking and FFT analysis can utilize whole signal from the IMU sensor output. By taking this proposal, not only MTC estimation but also time-series data for toe clearance become possible. As a result, the RMSE for MTC was 2.70 mm, which indicated the possibility of estimating the time change of toe clearance from the thigh and shin IMU sensors. When only the MTC was used for teaching data, the RMSE of the MTC was 2.41 mm.
