Effect of Mechanically Passive, Wearable Shoulder Exoskeletons on Muscle Output During Dynamic Upper Extremity Movements: A Computational Simulation Study

Abstract

Wearable passive (ie, spring powered) shoulder exoskeletons could reduce muscle output during motor tasks to help prevent or treat shoulder musculoskeletal disorders. However, most wearable passive shoulder exoskeletons have been designed and evaluated for static tasks, so it is unclear how they affect muscle output during dynamic tasks. The authors used a musculoskeletal model and Computed Muscle Control optimization to estimate muscle output with and without a wearable passive shoulder exoskeleton during 2 simulated dynamic tasks: abduction and upward reach. To an existing upper extremity musculoskeletal model, the authors added an exoskeleton model with 3-dimensional representations of the exoskeleton components, including a spring, cam wheel, force-transmitting shoulder cable, and wrapping surfaces that permitted the shoulder cable to wrap over the shoulder. The exoskeleton reduced net muscle-generated moments in positive shoulder elevation by 28% and 62% during the abduction and upward reach, respectively. However, muscle outputs (joint moments and muscle effort) were higher with the exoskeleton than without at some points of the movement. Muscle output was higher with the exoskeleton because the exoskeleton moment opposed the muscle-generated moment in some postures. The results of this study highlight the importance of evaluating muscle output for passive exoskeletons designed to support dynamic movements to ensure that the exoskeletons assist, rather than impede, movement.

Design and Performance Evaluation of a Wearable Passive Cable-driven Shoulder Exoskeleton

“…The WPCSE even reduced the activity of the trapezius and infraspinatus muscles. This result, promisingly, contrasts our previous computational study, which predicted higher activity in several muscles during negative shoulder elevation [27]. One possible explanation is that, during such movements, the WPCSE offloads muscles that contribute to positive shoulder elevation that are eccentrically contracted to control the shoulder's angular velocity.…”

“…We first determined the form of ℎ( ) in (5) to calculate the final geometry of the cam. The value of was set to 0.06 , which we estimated from a computational musculoskeletal model [27]. The cable tension force was calculated by dividing by the radius of the pulley which is attached to the output spool.…”

Methodologies for evaluating exoskeletons with industrial applications

“…Occasionally, work environments were simulated with real working tools, objects, or motion scenarios. Four studies evaluated exoskeletal prototypes in modelled environments [Aoustin and Formalskii 2018, Nelson et al 2020, Pan et al 2014, Yan et al 2018] without any sample. Five studies were performed in field analysing subjective opinions [Smets 2019, Hensel andKeil 2019] combined with muscular activity [Gillette and Stephenson 2019], range of motion [Iranzo et al 2020], as well as movement patterns and simulated compression forces [Graham et al 2009].…”

“…The analysis of movement patterns (e.g., recorded with force sensors [Baser et al 2019], inertial measurement units [Maurice et al 2020], video camera [Wang et al 2018], optical marker system [d 'Elia et al 2017]) focused mainly on kinematic aspects or working speed comparisons. Modelling (e.g., musculoskeletal simulations [Blanco et al 2019, Nelson et al 2020, Weston et al 2018, mathematical/numerical calculations [Aoustin and Formalskii 2018, Han et al 2020, Pan et al 2014] addressed mainly the system's support as well as the movability and motions during its use. Occasionally, input data from sample's anthropometry or force sensors are also considered.…”

Design and Preliminary Evaluation of a Wearable Passive Cam-Based Shoulder Exoskeleton

“…We first determined the form of ( ) in ( 5) to calculate the final geometry of the cam. The value of was set to 0.06 , which we estimated from a computational musculoskeletal model [32]. The cable tension force was calculated by dividing by the radius of the pulley which is attached to the output spool.…”