Adaptive pattern-alignment algorithm for gait phase estimation framework of soft wearable robotic suits

Accurate gait phase estimation is essential for synchronizing wearable robots with human motion. However, existing methods often rely on rigid sensors or computationally intensive models, which limit their applicability to fully soft wearable systems. This study presents a real-time gait phase estimation framework that integrates an embroidered stretch sensor with an Adaptive Pattern-Alignment Algorithm for soft wearable suits. The soft sensor was designed for reliable strain measurement and seamless integration into fabric-based garments, providing stable input signals for adaptive gait phase estimation. Feature attribution techniques–SHapley Additive exPlanations and Integrated Gradients–are employed to identify the most informative sensor placements. Adaptive normalization, combined with Dynamic Time Warping, standardizes sensor signals across different body morphologies and locomotion speeds. The proposed pattern-alignment algorithm achieved an average RMSE of 3.35% across 10 participants and maintained real-time performance with an average processing time of 0.046 ms per gait cycle, demonstrating its suitability for wearable robotic applications.