Hip exoskeleton assisted speed method for fast walking

Abstract

Fast walking, as a locomotor modality between conventional walking and running, presents distinct biomechanical demands characterized by a pronounced surge in hip joint torque and dynamic compression of the gait cycle. However, existing lower-limb exoskeletons commonly suffer from delayed intention recognition under continuously varying speeds and the inability of fixed-parameter assistance profiles to accommodate both individual differences and continuous speed transitions. To address these challenges, this thesis designs a lightweight rigid hip exoskeleton system and proposes a personalized, speed-adaptive assistance control strategy based on an asymmetric Gaussian curve and neural network, aiming to achieve efficient and safe augmentation of human fast walking. The performance verification experiments demonstrate that the proposed strategy can automatically generate assistance profiles that conform to individual gait characteristics based on the wearer’s physical traits. In variable speed walking tests, the personalized assistance mode can reduce the net metabolic rate by an average of 10.3% compared with the no-exoskeleton condition, while maintaining accurate IMU-based gait detection and smooth, mutation-free torque delivery. This study provides a feasible technical solution for hip exoskeleton systems intended for daily fast walking assistance.