Optimization and evaluation of lower limb exoskeleton based on human-exo coupling dynamics

The rapid advancement of lower limb exoskeleton technology has created new opportunities to enhance human locomotion and reduce operational fatigue. In our study, the in-depth investigation of human-exoskeleton coupling dynamics has been conducted and exoskeleton performance by evaluating assistive efficiency has been optimized. First, we propose a linear human-exoskeleton coupling model that integrates electromyography (EMG) features and human physiological parameters to predict coupling parameters using a CNN-LSTM neural network. On this basis, we refine the coupling model to a nonlinear framework, analyzing the effects of coupling position, looseness, and other influencing factors on the coupling parameters. Additionally, we propose a regression model for predictive analysis. To further explore human-exoskeleton interaction, we extract a dynamic physical model of coupling forces using a sparse regression algorithm. Additionally, we propose a mechanics-based ground reaction force calculation method and establish a human-exoskeleton coupling dynamic model for walking gait. The inertia parameters of both the exoskeleton and human body are experimentally identified to enhance model accuracy. Finally, to maximize the assistive effect of the lower limb exoskeleton, we introduce an optimization framework based on human energy efficiency principles. A set of comprehensive evaluation metrics, incorporating both joint and muscle dynamics, is established as a reference for exoskeleton performance evaluation.