Estimation of Gait Parameters From Trunk Movement Measured by Doppler Radar
Kenshi Saho, Keitaro Shioiri, Shoma Kudo, Masahiro Fujimoto
Abstract
This study proposes a novel method to estimate biomechanical gait parameters using the monostatic continuous wave Doppler radar data corresponding to trunk movement; the use of trunk movement leads to stable gait parameter estimations when compared to the conventional method which is based on leg movements because the received powers of the trunk echoes are larger and more stable than those of the legs. The proposed gait parameter estimation method employs trunk accelerations extracted from radar spectrograms. The accuracies of the gait parameters, such as the step length, swing time, and stance time, estimated using the proposed methods were evaluated by comparing them with the reference motion capture data. We experimentally demonstrated that the proposed trunk-based method estimated gait parameters with accuracy similar to that of the conventional leg-based method. Additionally, the proposed trunk-based method was able to estimate the swing and stance times, whereas the leg-based method failed to estimate them owing to instability of the ankle echoes. Furthermore, the combined method of the trunk and toe data achieved better accuracy in the estimation of swing and stance times. This study is the first to use data on trunk movements (not leg movements) for temporal gait parameter estimation using radar and demonstrated its practicality for realistic situations in clinical and daily gait assessment to grasp the health status of individuals such as risks of future falls and cognitive impairments.