A Data-Driven Personalized Gait Generation Method for Ankle Dorsiflexion Assistance on Soft Exosuit

Soft ankle exosuits have become an effective method for helping foot drop patients improve their gait. However, existing ankle dorsiflexion assistance strategies do not sufficiently consider individual differences. To address this issue, we propose a data-driven personalized gait generation method to maximize the match between the exosuit and ankle dorsiflexion of different individuals. In this study, we first model the ankle dorsiflexion gait based on ankle dynamics. Then, we build a classification model to map kinematic data to muscle activation. Finally, we achieve an online personalized generation of ankle dorsiflexion gait based on the covariance matrix adaptation evolution strategy. The personalized gait generation method proposed in this paper could reduce the muscle activation of the tibialis anterior during treadmill at a speed of 1.25 m/s. Compared to generic assistance, the personalized method achieves an average reduction of 31.5% in muscle activation, representing a 4.5% improvement. This finding validates the effectiveness of the personalized method and offers a tailored solution for ankle dorsiflexion assistance.