Research Article: Altering gait variability with an ankle exoskeleton

Date Published: October 24, 2018

Publisher: Public Library of Science

Author(s): Prokopios Antonellis, Samuel Galle, Dirk De Clercq, Philippe Malcolm, Alena Grabowski.


Exoskeletons can influence human gait. A healthy gait is characterized by a certain amount of variability compared to a non-healthy gait that has more inherent variability; however which exoskeleton assistance parameters are necessary to avoid increasing gait variability or to potentially lower gait variability below that of unassisted walking are unknown. This study investigated the interaction effects of exoskeleton timing and power on gait variability. Ten healthy participants walked on a treadmill with bilateral ankle-foot exoskeletons under ten conditions with different timing (varied from 36% to 54% of the stride) and power (varied from 0.2 to 0.5 W∙kg-1) combinations. We used the largest Lyapunov exponent (LyE) and maximum Floquet multiplier (FM) to evaluate the stride-to-stride fluctuations of the kinematic time series. We found the lowest LyE at the ankle and a significant reduction versus powered-off with exoskeleton power (summed for both legs) of 0.45 W∙kg-1 and actuation timing at 48% of the stride cycle. At the knee, a significant positive effect of power and a negative interaction effect of power and timing were found for LyE. We found significant positive interaction effects of the square of timing and power for LyE at the knee and hip joints. In contrast, the FM at the ankle increased with increasing power and later timing. We found a significant negative effect of power and a positive interaction effect of power and timing for FM at the knee and no significant effects of any of the exoskeleton parameters for FM at the hip. The ability of the exoskeleton to reduce the LyE at the ankle joint offers new possibilities in terms of altering gait variability, which could have applications for using exoskeletons as rehabilitation devices. Further efforts could examine if it is possible to simultaneously reduce the LyE and FM at one or more lower limb joints.

Partial Text

Walking is a fundamental skill and the prime gait of humans. When we walk, our footprints (e.g., those noticed in the sand or snow) never repeat themselves exactly, demonstrating the step-to-step variability in a continuous cycle of movement. This variability is inherent within all biological systems. Variability can be defined as the variations that occur in motor performance across multiple repetitions of a task over time. Traditional linear measures such as standard deviation and coefficient of variation can be used to assess variability. Although these measures can indicate the magnitude of variability at certain time occurrences, the temporal evolution of movement patterns is ignored. Gait parameters are often treated algorithmically (i.e., smoothing, differentiation and normalization) to provide a mean picture of the individual’s movement that distorts the temporal structure of variability [1]. In contrast, measures from nonlinear dynamics examine how neuromuscular behavior changes over time and provide information about the organization or structure of gait [1,2].

This study investigated the influence of exoskeleton power and the interaction with timing on gait variability. We found that actuation from an ankle exoskeleton can alter gait dynamics. The LyE showed that increased exoskeleton power reduced gait variability at the ankle compared to powered-off, but the FM showed the opposite effect. More specific, the Late-High condition with bilateral exoskeleton power of 0.45 W∙kg-1 and actuation timing at 48% resulted in the lowest LyE of all powered conditions and a lower LyE than walking with the exoskeleton without assistance from the pneumatic muscles (powered-off). Conversely, the FM showed that higher exoskeleton power and late timing increased gait variability at the ankle, and the FM was highest with bilateral exoskeleton power of 0.37 W∙kg-1 and actuation timing of 42%. At the knee, we found that the LyE was lowest in the powered-off condition, and the Early-High condition resulted in the highest LyE. No effects of actuation timing or power were found at the hip for the LyE or at the hip and knee for the FM.

This study examined the effects of ankle exoskeleton power and the interaction with actuation timing on gait variability. During exoskeleton walking with pneumatic muscle assistance, individuals ambulate with reduced LyE at the ankle joint, which is one measure of gait variability. We found the lowest LyE for the ankle with actuation timing at 48% of the stride and average exoskeleton power at approximately 0.45 W∙kg-1. No statistically significant differences in gait variability were reported at the ankle between the powered exoskeleton conditions and walking without the exoskeleton. On the other hand, the FM indicated that higher exoskeleton power increased gait variability at the ankle, with a bilateral exoskeleton power of 0.37 W∙kg-1 corresponding to the highest FM. Additionally, exoskeleton power and actuation timing had effects on the LyE and FM at the knee, and the LyE at the hip. In contrast, exoskeleton power and actuation timing did not significantly affect the FM at the hip. Our study results provide new insights into the human response to external assistance in terms of altering and possibly reducing gait variability, which may benefit future studies focused on designing and controlling lower limb exoskeleton devices. This needs to be further tested for the FM via experiments possibly using diverse and pathological populations. Future work should investigate whether this approach provides similar changes and potentially even improvements in patient populations with reduced exercise tolerance, and examine the sensitivity and specificity of the two nonlinear methods used in this study to aid the design of improved exoskeleton devices for rehabilitation and clinical use.




0 0 vote
Article Rating
Notify of
Inline Feedbacks
View all comments