Text

Introduction: Evaluating the biomechanical impact of exoskeletons is a key factor in accelerating their use in MSD prevention. Such systems transfer the load from one part of the body to another, inducing beneficial and undesirable effects. To assess these effects, it is necessary to integrate the forces generated at the interface between the operator and the system into a biomechanical analysis of the task, either by modeling the contact or by indirectly measuring this action. We propose a methodology for evaluating the effects of a passive shoulder-assist exoskeleton through indirect measurement of forces and pressure centers at the interfaces between the operator and the system.

Methods: We evaluated the Ottobock Shoulder (Ottobock SE & Co. KGaA), a passive exoskeleton of shoulder flexion maximal support torque of 8 Nm. A 6-axis force transducer (ATI Industrial Automation, 2040 Hz) was integrated into its main dorsal arm. We equipped one subject with the exoskeleton set at 75% of its maximal assistance, reflective markers, to capture his movement (Qualysis,16 cameras, 120 Hz) and 2 pressure maps (Xsensor, 60 Hz) on the belt and armband, to measure the pressure centers at the interfaces. The subject performed shoulder sagittal flexion/extension movements at free speed. The data was used to perform inverse kinematics and inverse dynamics in OpenSim 4.5, using a full-body model. The actions measured by the force transducer were applied to the sacrum and arm respectively, at the centers of pressure estimated by the maps. Joint torques at the shoulder and L5/S1 were compared without and with exoskeleton actions.

Results: Shoulder joint torques varied from a maximum of 8.7 Nm to 4.6Nm in flexion, -7.3 Nm to 0 Nm in extension, 13.8 Nm to 0.7 Nm in adduction, -12.9 Nm to -2.1 Nm in abduction, 0 Nm to 2 Nm in internal rotation, and -2.4 Nm to -1 Nm in external rotation, without and with the exoskeleton respectively. L5/S1 joint torques varied from a maximum of 1.5 Nm to 0.6 Nm in flexion, -1.6 to -0.4 Nm in extension, 5.5 Nm to 0.5 Nm in lateral flexion, 0.6 Nm to 0 Nm in lateral extension, 0.2 Nm to 0 Nm in internal rotation and -0.7 Nm to -0.1 Nm in external rotation, without and with the exoskeleton respectively.

Discussion: We observed a decrease in maximal shoulder flexion torque (-47%) and an in maximal shoulder abduction torque (-95%), demonstrating joint compensation for the action of the exoskeleton. This may imply a lesser demand on the deltoid anterior and pectoralis major. We also observed a decrease in most L5/S1 moments. Thus, the presence of the exoskeleton may lead to beneficial effects on both joints. These preliminary results need to be investigated on a larger population with more detailed estimates of muscle forces to investigate the physical changes related to the exoskeleton action.

Conclusion: We carried out an initial assessment of the biomechanical impact of a passive shoulder-assist exoskeleton, using indirect measurements of forces at the interfaces. The exoskeleton effectively relieved shoulder flexion. A study on a larger cohort including musculoskeletal analysis is required to qualify these initial results.