Frequency-estimation-based Thigh Angle Prediction in Level-ground Walking Using IMUs

We study thigh angle prediction in level-ground walking using inertial measurement units (IMUs). Our approach take advantage of the periodicity of thigh movements during level-ground walking to identify a new method, which formulates the thigh angular data into a multi-tone frequency model and reconstruct data for prediction based on the frequency estimation results. To test the accuracy of our proposed algorithm, we collect data from one thigh and take it as a reference for the prediction result of the other's. Furthermore, both intra-subject and inter-subject performances of our prediction algorithm are validated through multiple datasets collected from ten different subjects.