Slow intermuscular oscillations and strategies to tango with a robot

Abstract

Steadiness and stiffness control failures are recurring problems and possibly a byproduct of "common drive", a nervous system process that activates or relaxes all muscles in synchrony. Slow correlated neural oscillations, a function of common drive, were investigated for improving concurrent activation control around elbow joint muscles with direct application to HRI. Experiments were designed, data collected from 80 healthy subjects. Many methods to assess and quantify common intermuscular oscillations were tested including event related coherence (amplitude and phase) and in-phase synchrony. Linear associations between system performance (neural or mechanical) and muscle correlated oscillations (amplitude and phase coherence) under different conditions (static, dynamic, and transient coactivation) were established. Multiple methods were studied to modulate or influence these associations such as repetition, and intervention (out-of-phase cocontraction, and single muscle habituation). Findings suggest that a proper HRI framework would benefit from a good grasp of task-specific demands as well as system hardware properties in addition to models of neural and mechanical output as a function of internal processes such as correlated oscillations.