Predictive Gait Simulations of Human Energy Optimization

Abstract

We previously demonstrated that humans can continuously adapt their gait to optimize energetic cost in real-time when wearing a lower-limb exoskeleton. Here, we aim to recreate this paradigm using predictive gait simulations to further investigate how the nervous system performs this optimization and how energy costs change locally. To match the real-world experiment, we modeled a knee-worn exoskeleton that applied resistive torques that were either proportional or inversely proportional to step frequency—decreasing or increasing the energy optimal step frequency, respectively. We solved simulations with and without the knee exoskeleton and with fixed and free step frequency. We were able to replicate the experiment, finding higher and lower optimal step frequencies than in the natural walking under each respective condition. Our simulated resistive torques and optimized objective function resembled the measured experimental resistive torque and metabolic energy landscape. Muscle metabolic power changed for individual muscles spanning all three joints and revealed distinct coordination strategies consistent with each exoskeleton controller condition.