Endpoint Upper Limb Stiffness Computation and Characterization in a Ball Bouncing Task Using EMGs

A key aspect to understand human dynamic control is to estimate their impedance in interaction with an environment. This work focuses on estimating the arm stiffness, modeled as a second-order spring-mass-damper system, during a ball-bouncing task involving intermittent interaction with the environment. Surface electromyography (sEMG) is employed in this study, enabling continuous estimation of muscle activity and stiffness using Hill-type muscle models on thirty-four participants. After a training phase, two experiments were performed, with and without haptic feedback, to provide insights on its effect on human control. This study analyzes different dynamic control strategies found during the experiment and the linked stiffness profiles. Results found two major strategies, one with a symmetrical force/position profile and one asymmetrical with a shorter upward arm's motion phase. The resulting stiffness profiles were closely related to those force/position profiles. The stiffness from the asymmetrical profile was higher due to more active control of the motion. However, both strategies were within the limits of the passive stability domain. The symmetrical profile performed better on the task, though both profiles had an increase in variability with the added haptic feedback.