Validating Model-Based Prediction Of Biological Knee Moment During Walking With An Exoskeleton in Crouch Gait: Potential Application for Exoskeleton Control.

Abstract

Advanced control strategies that can adjust assistance based volitional effort from the user may be beneficial for deploying exoskeletons for overground gait training in ambulatory populations, such as children with cerebral palsy (CP). In this study, we evaluate the ability to predict biological knee moment during stance phase of walking with an exoskeleton in two children subjects with crouch gait from CP. The predictive model characterized the knee as a rotational spring with the addition of correction factors at knee extensor moment extrema to predict the instantaneous knee moment profile from the knee angle. Our model prediction performance was comparable to previous studies for weight acceptance (WA) and mid-stance (MS) phases in both assisted (Assist) and non-assisted (Zero) modes based on normalized root mean square error (RMSE), demonstrating the feasibility of joint moment estimation during exoskeleton walking. RMSE was highest in late stance phase, likely due to the non-linear knee stiffness exhibited during this phase in one participant. Overall, our results support real-time implementation of the joint moment prediction model for control of exoskeleton knee extension assistance in children with CP.