Estimation of Step Length and Gait Asymmetry Using Wearable Inertial Sensors

Abstract

Spatial-asymmetry of gait is important to reflect the different walking abilities between human left side and right side. Most of the studies estimated spatial-asymmetry of gait by the laboratory-based motion capture system. However, a little work has been done by using a wearable inertial measurement unit (IMU), despite these sensors have a smaller size and better portability. In this paper, a simple geometrical method was developed to estimate step length and its asymmetry between two sides with measurements from four IMUs attached on lower limbs. The step length asymmetry was used to estimate the spatial-asymmetry of gait. Overground walking experiments with ten healthy subjects and five subjects with gait impairment (four post stroke and one sciatic nerve injury) under different walking conditions were performed to evaluate the performance of the proposed method. For healthy subjects, the root-mean-square error (RMSE) in step length estimation is 5.4 cm on average and in gait asymmetry estimation is 3.0%. For subjects with gait impairment, the RMSE in step length estimation is 9.6 cm and in gait asymmetry estimation is 14.7%. The algorithm can be easily implemented in real time for potential applications, such as wearable assistive device control or biofeedback gait re-training.