A Novel Wavelet-Based Gait Segmentation Method for a Portable Hip Exoskeleton

For a lower limb exoskeleton, accurate and continuous estimation of the gait phase in real time is a fundamental requirement to provide a well-tailored assistive action at the proper time in the gait cycle (GC). This article presents a new gait phase estimator for a portable hip exoskeleton based on discrete wavelet transform (DWT) and adaptive oscillators. The algorithm is capable of continuously tracking the gait phase and identifying the relevant biomechanical gait events online, i.e., heel strike (HS) and toe-off (TO). The proposed method exploits only the hip joint angle signals measured by hip encoders, avoiding the need to use additional sensors to those already built in the exoskeleton. The novel phase estimator has been benchmarked against a state-of-the-art method, based on the maximum flexion angle (MFA), with pressure-sensitive insoles used as the reference ground truth for the event detection. To validate the method, two experimental activities were carried out. Experiments conducted with eight healthy subjects walking on a treadmill at different speeds, with and without hip assistance, demonstrated that the DWT-based method outperformed the MFA method in all operative conditions, reducing the rms of the phase reset error by 64.0% in assistive mode, and identifying the HS and TO events with low delay (0.4% and 1.1% GC, respectively, for HS and TO). Experiments carried out with three transfemoral amputees showed similar performance, paving the way for clinical applications of the method.