Predictive simulation framework for replicating energy reduction trends and joint kinematic adaptations in walking with powered ankle exoskeletons

Exoskeletons are useful for performance augmentation and rehabilitation, yet the design and control optimization of these devices are challenging, iterative, and laborious. Predictive simulations present an opportunity to complement experimental approaches in addressing these challenges. In this study, we investigated the ability of a data-tracking predictive simulation framework to replicate the documented trends in energy reductions and joint kinematic adaptations observed in the literature during walking assisted by powered ankle exoskeletons. We formulated an optimal control problem utilizing a musculoskeletal model, aiming to determine the states and controls that minimize a weighted combination of tracking error and muscular effort while considering musculoskeletal dynamics and task constraints. Two scenarios were examined by varying the weight of the muscular effort term in the cost function: preference for preserving normal kinematics and preference for reduced muscular effort. The reference data for tracking was obtained from the mean experimental data of walking without an exoskeleton. The optimal control problem was transformed into a nonlinear programming problem using direct collocation and subsequently solved. Our simulation framework demonstrated the ability to predict the optimal assistance onset time, as well as the effects of different assistance levels and onset timings on the metabolic rate. It also predicted increased ankle plantarflexion with increased assistance, consistent with experimental findings, and suggested that allowing for kinematic adaptations would increase energy reduction. Hence, this framework, with further improvements, has the potential to be used as a tool to optimize design and control strategies for exoskeletons.