A Model and Experiment of Human Control of Lateral Balance During Walking

Abstract

An essential component of human walking is the control of balance. Inadequate balance control can lead to falls and injuries, which are major concerns for older adults and those with walking impediments. Previous examinations of balance control during walking have largely been focused on stepping strategies, such as step placement control and stance ankle torque, that directly control the location of the center of pressure relative to the body center of mass. However, a relatively unappreciated approach is to indirectly affect the center of mass through inertial balancing strategies, where parts of the body like the trunk and arms are moved to induce a stabilizing reaction force. Inertial balance can significantly contribute to balance recovery during standing and on constrained surfaces such a tight-rope. However, the contribution of these inertial controls has scarcely been examined during walking, and particularly for lateral stability, which is thought to require more active control than forward stability. I intend to bridge this gap by demonstrating how inertial strategies can contribute to lateral balance control in response to perturbations during walking.

This work consists of a model of human walking and an experiment. A simple dynamical model is used to predict how both stepping and inertial strategies may be controlled to achieve lateral balance. The model walks in 3D largely through passive dynamics but also has active control for key degrees of freedom such as push-off, step placement, and trunk roll. I show how stepping and inertial balance controls may be designed using a once-per-step hybrid control scheme, and demonstrate how the two can be coupled operationally to counteract against lateral balance disturbances.

The experiment consists of lateral perturbations applied to human subjects as they walk. I designed a force feedback device to apply perturbations to a freely moving subject on an instrumented treadmill, with ability to modulate the timing and amount of force applied. The experiments were designed to induce a variety of balance strategies, both while walking normally and with constraints on available step placement, similar to a narrow walkway. This demonstrates how humans modulate and select from a continuum of control between stepping and inertial strategies.

There were several model predictions that were validated by the experiment. One is that perturbations during different parts of the stride call for different amounts of foot placement, due in part to the time available before the next footfall. The model also predicts that multiple balance strategies should contribute simultaneously, and particularly so with narrow walking constraints. Experiments show that humans do modulate their foot placement according to perturbation direction and timing, with greater placement for earlier perturbations in a step. However, stance ankle torque and inertial balance strategies also contribute to a smaller degree throughout the stride, affecting the rate and timing of falling, and thus the amount of foot placement needed. These effects were amplified in the presence of narrow stepping constraints.

This dissertation shows how balance during walking is controllable by multiple strategies and quantifies how humans employ them in combination. Inertial balance strategies are employed in many cases, particularly when there is considerable time before the next footfall. This work also demonstrates how computational models can explain active balance control, and how humans can flexibly combine multiple strategies to accommodate various challenges and constraints.