Abstract
Embedded systems in exoskeletons and wearable rehabilitation devices require efficient monitoring of human gait with stringent demands for real-time and reliable performance. Gait trajectory prediction enables full-process perception of future gait patterns. The template matching approach is a promising algorithm for gait trajectory prediction. However, it has challenges in terms of template selection and dynamic stability. We developed DTW-KM Template Selection Method using the DTW distance matrix and K-means to find trajectories with generality and diversity as candidate templates. In addition, we use a matching method combining soft constraints and quadratic weighting, named ScW Template Matching Method. it can balance timeliness, stability and flexibility, and suppress gait phase and trajectory oscillations. Preliminary experimental results show that our method achieves stable prediction of 0.5--1s gait using efficient inference on embedded devices.
Funding Information
  • Shenzhen Municipal Natural Science Foundation and Shenzhen Science and Technology Innovation Committee (KCXFZ202002011010487)
  • Shenzhen Municipal Natural Science Foundation (WDZC20200818121348001)