The Benefit of Combining Neuronal Feedback and Feed-Forward Control for Robustness in Step Down Perturbations of Simulated Human Walking Depends on the Muscle Function

Abstract

It is often assumed that the spinal control of human locomotion combines feed-forward central pattern generation with sensory feedback via muscle reflexes. However, the actual contribution of each component to the generation and stabilization of gait is not well understood, as direct experimental evidence for either is difficult to obtain. We here investigate the relative contribution of the two components to gait stability in a simulation model of human walking. Specifically, we hypothesize that a simple linear combination of feedback and feed-forward control at the level of the spinal cord improves the reaction to unexpected step down perturbations. In previous work, we found preliminary evidence supporting this hypothesis when studying a very reduced model of rebounding behaviors. In the present work, we investigate if the evidence extends to a more realistic model of human walking. We revisit a model that has previously been published and relies on spinal feedback control to generate walking. We extend the control of this model with a feed-forward muscle activation pattern. The feed-forward pattern is recorded from the unperturbed feedback control output. We find that the improvement in the robustness of the walking model with respect to step down perturbations depends on the ratio between the two strategies and on the muscle to which they are applied. The results suggest that combining feed-forward and feedback control is not guaranteed to improve locomotion, as the beneficial effects are dependent on the muscle and its function during walking.

Measuring Gait Stability in People with Multiple Sclerosis Using Different Sensor Locations and Time Scales

“…Perturbations seem to be better compensated in the healthy comparison group on a longer time scale based on trunk movements and on a shorter time scale (almost instantaneously) according to the foot kinematics. Probably, some mechanisms to compensate perturbations occur nearly instantly or in a short period within the lower extremities, for example, when preparing the leg for ground contact by an adequate adjustment of muscle (pre-) activation [32,33] or joint stiffness [34]. On the other hand, some trunk-based mechanisms might need more time to affect stability, e.g., recovery movements from perturbations due to armswing [35] or directing the ground reaction forces to an intersection point above the center of mass [36,37].…”

Simulating the effect of ankle plantarflexion and inversion-eversion exoskeleton torques on center of mass kinematics during walking

“…Previous studies have used simulation to study walking balance using a variety of approaches [14,[21][22][23][24][25][26][27][28][29][30][31][32][33][34][35]. Simplified biped models have been used to study how modulating push-off [28,32] and joint speed [22] can help stabilize walking.…”

“…Simplified biped models have been used to study how modulating push-off [28, 32] and joint speed [22] can help stabilize walking. Several studies have simulated how humans maintain walking balance using models of muscle-reflex through feedback control [26, 27, 33, 34] based on the model of Geyer and Herr (2010) [36], and others have incorporated feed-forward control strategies to model the contribution of central pattern generators to walking balance [21, 23, 30]. Three-dimensional musculoskeletal simulations have revealed the contributions of muscles to medio-lateral ground reaction forces [14, 24] and foot placement strategies after perturbation [29, 35].…”

A controller perspective on biological gait control: Reflexes and central pattern generators

“…Initially it was claimed that this type of feedback was sufficient to evoke locomotion, even in the absence of a CPG, but more recent work showed that the addition of CPG to this model greatly improved its performance (Dzeladini, van den Kieboom, & Ijspeert, 2014;Haeufle, Schmortte, Geyer, Müller, & Schmitt, 2018). More recently, a combination of CPGs with bioinspired sensory feedback was employed to make a complex 3D biped robot (COMAN) walk stably with a straightforward speed control (Van der Noot, Ijspeert, & Ronsse, 2018a; Van der Noot, Ijspeert, & Ronsse, 2018b).…”