Lower limb biomechanical analysis during an unanticipated step on a bump reveals specific adaptations of walking on uneven terrains.

Although it is clear that walking over different irregular terrain is associated with altered biomechanics, there is little understanding of how we quickly adapt to unexpected variations in terrain. This study aims to investigate which adaptive strategies humans adopt when performing an unanticipated step on an irregular surface, specifically a small bump. Nine healthy male participants walked at their preferred walking speed along a straight walkway during five conditions: four involving unanticipated bumps of two different heights, and one level walking condition. Muscle activation of eight lower limb muscles and three-dimensional gait analysis were evaluated during these testing conditions. Two distinct adaptive strategies were found, which involved no significant change in total lower limb mechanical work or walking speed. An ankle-based strategy was adopted when stepping on a bump with the forefoot, whereas a hip-based strategy was preferred when stepping with the rearfoot. These strategies were driven by a higher activation of the plantarflexor muscles (6-51%), which generated a higher ankle joint moment during the forefoot conditions and by a higher activation of the quadriceps muscles (36-93%), which produced a higher knee joint moment and hip joint power during the rearfoot conditions. These findings provide insights into how humans quickly react to unexpected events and could be used to inform the design of adaptive controllers for wearable robots intended for use in unstructured environments that can provide optimal assistance to the different lower limb joints.

Human-in-the-loop Bayesian optimization of wearable device parameters.

The increasing capabilities of exoskeletons and powered prosthetics for walking assistance have paved the way for more sophisticated and individualized control strategies. In response to this opportunity, recent work on human-in-the-loop optimization has considered the problem of automatically tuning control parameters based on realtime physiological measurements. However, the common use of metabolic cost as a performance metric creates significant experimental challenges due to its long measurement times and low signal-to-noise ratio. We evaluate the use of Bayesian optimization-a family of sample-efficient, noise-tolerant, and global optimization methods-for quickly identifying near-optimal control parameters. To manage experimental complexity and provide comparisons against related work, we consider the task of minimizing metabolic cost by optimizing walking step frequencies in unaided human subjects. Compared to an existing approach based on gradient descent, Bayesian optimization identified a near-optimal step frequency with a faster time to convergence (12 minutes, p < 0.01), smaller inter-subject variability in convergence time (± 2 minutes, p < 0.01), and lower overall energy expenditure (p < 0.01).