Emg-informed Computed Muscle Control for Dynamic Simulations of Movement

INTRODUCTION Computed Muscle Control (CMC) is an efficient method for estimating muscle activations and forces from kinematic and ground reaction force data [1]. However, it can fail to replicate some muscle activation characteristics found in electromographic (EMG) recordings. For instance, muscles often activate in anticipation of events such as heel strike, and these anticiptory responses are not predicted by CMC. Also, when joint accelerations are high, for instance at the ankle during heel strike or toe-off, CMC will often predict co-contraction strategies to track the kinematics when gravity or inertia were actually providing the necessary forces. Previous efforts to reconcile these differences have attempted to implement constraints on muscle activations based on recorded EMG data [2]. This can cause simulations to fail if the constraint criteria cannot be met or can result in unwanted effects on unconstrained muscles. We present here a modified version of CMC that better reproduces measured EMG data without constraining muscle activations.