Step-to-step variations in human running reveal how humans run without falling

Abstract

Humans can run without falling down, usually despite uneven terrain or occasional pushes. Even without such external perturbations, intrinsic sources like sensorimotor noise perturb the running motion incessantly, making each step variable. Here, using simple and generalizable models, we show that even such small step-to-step variability contains considerable information about strategies used to run stably. Deviations in the center of mass motion predict the corrective strategies during the next stance, well in advance of foot touchdown. Horizontal motion is stabilized by total leg impulse modulations, whereas the vertical motion is stabilized by differentially modulating the impulse within stance. We implement these human-derived control strategies on a simple computational biped, showing that it runs stably for hundreds of steps despite incessant noise-like perturbations or larger discrete perturbations. This running controller derived from natural variability echoes behaviors observed in previous animal and robot studies.

A controller for walking derived from how humans recover from perturbations

“…To meaningfully compare the inferred feedback gains from different subjects or subject populations or experiments, we would first need to accurately characterize the trial to trial variability of such gains for individual subjects: repeating over different days and repeating with different experimenters with similar marker-placement instructions. One potential way to reduce variability due to experimental practice or marker placement may be to obtain CoM position and velocity estimates by combining integrated ground reaction force data and marker data with an appropriate Kalman-like filter to avoid drift [33,44].…”

Peer Review #1 of "The effect of external lateral stabilization on the use of foot placement to control mediolateral stability in walking and running (v0.3)"

“…Similar to previous studies, which reported that 50-84% of ML foot placement variance can be explained by ML trunk, ML pelvis, or ML whole-body CoM state during walking [14][15][16], our results indicated high predictive ability of ML trunk CoM state on subsequent ML foot placement, with R 2 ranging between 52-85% during the entire swing phase in walking. Recently, Seethapathi and Sirinivasan [8] reported that ML foot Manuscript to be reviewed movements to movements of the upper body. Although the results of current study cannot answer the question whether active control or passive coupling is the underlying cause of this correlation, active control of ML stability through foot placement is supported by studies on the effects of sensory illusions induced by vibration [17], or visual perturbations [18] on this correlation, and by studies that have related ML foot placement to swing phase muscle activity [19].…”

Dynamical determinants of different spine movements and gait speeds in rotary and transverse gallops

“…To overcome the limitations of the observational approach in animal locomotion, modeling approaches have attracted recent research attention ( Bertram and Gutmann, 2009 ; Alexander, 1988 ; Marques et al, 2014 ; Markowitz and Herr, 2016 ; Swanstrom et al, 2005 ; Aoi et al, 2017 ; Usherwood and Davies, 2017 ; Ambe et al, 2018 ; Fujiki et al, 2018 ; Aoi et al, 2019 ; Toeda et al, 2019 ). Because the essential contribution of the legs can be represented by springs, the spring-loaded inverted pendulum (SLIP) model was developed to investigate animal locomotion mechanism from a dynamic viewpoint, particularly for human running and walking ( Blickhan, 1989 ; McMahon and Cheng, 1990 ; Seyfarth et al, 2002 ; Geyer et al, 2005, 2006 ; Srinivasan and Holmes, 2008 ; Lipfert et al, 2012 ; Seethapathi and Srinivasan, 2019 ). The SLIP model has been improved for examining gait in quadrupeds ( Full and Koditschek, 1999 ; Blickhan and Full, 1993 ; Farley et al, 1993 ; Tanase et al, 2015 ) to clarify the common and unique principles between animal gaits.…”