Emulating Balance Control Observed in Human Test Subjects with a Neural Network

Abstract

Human balance is likely achieved using many concurrent control loops that combine to react to changes in environment, posture, center of mass and other factors affecting stability. Though numerous engineering models of human balance control have been tested, no models of how these controllers might operate within the nervous system have yet been established. We have developed such a neural model, focusing on a proprioceptive feedback loop. For this model, angular position is measured at the ankle and corrective torque is applied about the joint to maintain a vertical orientation. We built a physical model of an upright human maintaining balance with an inverted pendulum actuated by a torque-control motor. We used an engineering control model for human balance to calculate the control parameters that will cause our physical model to have the same dynamics as human test subject data collected on a tilting platform. We reconstruct this controller in a neural network and compare performance between the neural and classical engineering models in experiment, demonstrating that the design tools in this paper can be used to emulate a classical controller using a neural network with relatively few free parameters.