Improving balance with less effort through faster-than-physiological ankle exoskeleton assistance during perturbed gait: Humans adapt fast with altered sensorimotor processing

Background: Supporting human balance is a crucial application of powered lower-limb exoskeletons, as proper balance control could improve user safety, reduce fall risk for populations with mobility impairments, and ultimately promote device adoption in the real world. Recent work suggests that exoskeleton balance assistance provided faster than the physiological response improves postural balance; however, this approach has not yet been investigated during gait. Thus, we aimed to test faster-than-physiological ankle exoskeleton balance assistance during perturbed gait while exploring other knowledge gaps regarding ankle exoskeleton balance control. Specifically, we investigated (1) the effects of assistance on balance performance and recovery effort, (2) the extent of adaptation to the assistance, and (3) sensorimotor processing with versus without the assistance. Methods: Thirteen adults without mobility impairments participated in an ankle exoskeleton balance study. They walked on a treadmill while wearing bilateral ankle exoskeletons and receiving unexpected forward pushes of three magnitudes from a robotic pusher device attached to their waist. Center-of-mass states, ankle moment and muscle activity, and spatiotemporal gait outcomes during push responses were assessed without balance assistance, during exposure to balance assistance for 180 pushes, and upon removal of the balance assistance. Results: With ankle exoskeleton balance assistance, participants responded to forward pushes with significantly improved balance performance (decreased center-of-mass states) and reduced recovery effort (decreased soleus muscle activity and ankle moment). Participants adapted quickly to the assistance by decreasing their ankle muscle activity and increasing their center-of-mass displacement, without aftereffects, when assistance was removed. Sensorimotor processing in response to forward pushes (relationship between center-of-mass state and resulting muscle activity) significantly differed with versus without the balance assistance. Conclusion: Faster-than-physiological ankle support may be a promising solution to improve the balance of exoskeleton users during gait, as participants exhibited better balance performance and reduced effort. Participants' quick adaptation without aftereffects provides insights into future exoskeleton training paradigms, and sensorimotor findings provide further understanding of how humans internalize the use of robotic support. Overall this work contributes to the integration of balance control into lower-limb exoskeletons, advancing toward the ultimate goal of a wearable robotic solution for individuals with balance impairments.