Step Length Estimation with Wearable Wrist Sensor using ANN.

Step Length is an important metric that can be used for the analysis and assessment of the gait. Proper dynamical models are not available in current literature associated with the wrist that can adequately determine the step length using recursive estimation techniques. This study presents a method to estimate the step length using angular velocity data from the wrist sensor. The technique maps the dynamical region corresponding to periods of activity of the gait manifested in angular velocity from the inertial measurement unit located at the wrist to that of the thigh using an artificial neural network, upon which an unscented Kalman filter is used to determine the horizontal position of the foot relative to the hip, and consequently, determine step length. The results for Step Length indicate an average accuracy of 81.8% and 91.1% for the young and elderly, respectively, when compared to a reference system, which, in our study, is data from a treadmill.