Adaptive Ankle Exoskeleton Control: Validation Across Diverse Walking Conditions

Abstract

Ankle exoskeletons hold potential to augment human walking ability, yet their use in free-living environments has been limited by the absence of practical and effective control strategies that can appropriately adapt to variable terrain. To address this challenge, we derived a novel analytical ankle joint moment estimation model using custom wearable sensors and developed an exoskeleton control scheme to adapt assistance proportional to the biological plantar-flexor moment in real-time for unimpaired individuals and individuals with disabilities who are able to ambulate independently. We validated the controller during level, 5◦ incline and decline walking, each at multiple speeds; stair ascent; stair descent; and 90◦ turning (88 ± 3% average accuracy, R = 0.96 ± 0.01 average correlation coefficient). This study demonstrates the ability of the controller to accurately adapt assistance in unimpaired individuals across a wide variety of walking conditions without the need for walking condition classification or real-time assessment of muscle activity. Clinical feasibility testing in four individuals with cerebral palsy suggests that this control method holds promise for incline and stair walking in individuals with mild-to-severe impairment. This ankle-moment-adaptive control system can be used to prescribe ankle exoskeleton assistance that adapts in real-time across the tested conditions.