Study of a Cable‐Driven Hip Swimming‐Assisted Exoskeleton Utilizing Adaptive Active Control Strategy

Abstract

ABSTRACTThe power‐assisted exoskeletons have a far‐reaching research significance. The adaptive active control strategy (AACS) of the exoskeleton allows the wearers to achieve personalized assistance tailored to their specific movement patterns and physical needs. However, due to the high requirements for waterproof design and the high difficulty in underwater AACS design, the researches on underwater power‐assisting exoskeletons are hardly to be seen. Therefore, this research proposes a cable‐driven hip swimming‐assisted exoskeleton (SAE) utilizing an AACS. SAE can generate adaptive power‐assisting guiding trajectories of thighs by underwater motion intention perception (UMIP) of the wearer. To achieve UMIP, a deep learning network is presented by combining conventional neural network, bidirectional long short‐term memory network, and attention mechanism (AM‐Dconvbilstm). SAE guides the wearer's thighs to move along the generated guiding trajectory by UMIP through the cables to conduct hip power assisting. The feasibility of the proposed AM‐Dconvbilstm on UMIP and the power‐assisting effect of designed SAE are tested on three elite swimmers by recording corresponding data from IMUs and a sports watch. According to the results of generated trajectories from SAE and actual trajectories from the wearer, the proposed AM‐Dconvbilstm model is verified to predict the wearer's hip motion intention accurately. The power‐assisting effect of SAE is also verified according to the sports watch's heart rate and average energy consumption. All the experiment results greatly validated the feasibility of the proposed AACS and designed SAE.