Evaluating Muscle Activation Models for Elbow Motion Estimation

Abstract

Adoption of wearable assistive technologies relies heavily on improvement of existing control system models. Knowing which models to use and how to improve them is difficult to determine due to the number of proposed solutions with relatively little broad comparisons. One type of these models, muscle activation models, describes the nonlinear relationship between neural inputs and mechanical activation of the muscle. Many muscle activation models can be found in the literature, but no comparison is available to guide the community on limitations and improvements. In this research, an EMG-driven elbow motion model is developed for the purpose of evaluating muscle activation models. Seven muscle activation models are used in an optimization procedure to determine which model has the best performance. Root mean square errors in muscle torque estimation range from 1.67–2.19 Nm on average over varying input trajectories. The computational resource demand was also measured during the optimization procedure, as it is an important aspect for determining if a model is feasible for use in a particular wearable assistive device. This study provides insight into the ability of these models to estimate elbow motion and the trade-off between estimation accuracy and computational demand.

A Mixed-Methods Approach to Evaluating the Internal Validity of the Reactive Strength Index

“…Actuation of tension within the muscle-tendon complex is realized from integration of active and passive elements. Hill-type models of muscle tension generally comprise an active contractile element and passive elastic elements arranged serially and in parallel [1,2,3]. Moreover, the mechanical behavior of the muscle-tendon complex is typically categorized into one of four principal actions: concentric, isometric, eccentric, and reactive [4].…”

“…Moreover, the mechanical behavior of the muscle-tendon complex is typically categorized into one of four principal actions: concentric, isometric, eccentric, and reactive [4]. Humans’ propensity for producing muscular strength and power during concentric, isometric, and eccentric actions is effectually described by the force-length and force-velocity dependencies of active and passive muscle-tendon elements [1,2,3]. Modelling the reactive capacity or strength of the muscle-tendon complex is more contemporary and has attracted considerable attention in recent literature.…”

Estimating the Effects of Awareness on Neck-Muscle Loading in Frontal Impacts with EMG and MC Sensors

“…As a non-invasive technique, surface EMG is widely applied, but it is also subject to numerous factors in the measuring and processing of EMG data that can compromise the outcome [ 26 , 27 , 28 , 29 , 30 ]. The post-processed and normalised EMG signal represents a time-dependent input parameter for muscle-activation dynamics that introduce a delay between the muscle’s electrical and mechanical onset, and non-linear relation between the EMG and the generated muscle force [ 31 , 32 ]. With the activation known, a Hill-type phenomenological model [ 33 ] is commonly used for the contraction dynamics to simulate the force generated by the muscle-tendon units.…”

Modeling of Motorized Orthosis Control

“…Tsai et al [16] carried out a comparison of upper-limb motion pattern recognition while using multichannel EMG signals from the shoulder and elbow during dynamic and isometric muscle contractions. Suberbiola et al [17] captured the biceps and triceps EMG signals and found more precise movement with the use of algorithms that are based on autoregressive models and artificial neural networks Recently, Desplenter and Trejos [18] have evaluated seven muscle activation models by means of an EMG-driven elbow motion model. Many differences were found between models suggesting that further evaluation and improvements of motion models is still necessary in order to increase the safety and feasibility of the wearable assistive devices.…”

“…The calculation of the variable L CE is achieved by means of Equations (13)-(16): (16) Once defined the muscle model, it is necessary to describe the dynamics of the arm by means of the Equation 17, as shown in Figure A5. The momentum of inertia J arm is obtained with the Equation (18) in order to achieve the arm dynamics.…”