Mechanical sensing of skeletal muscle contraction and fatigue for physical human-robot interaction

Abstract

This dissertation describes a mechanical sensing system for muscle contraction. A resonance-based active muscle stiffness sensor (aMSS) is developed for the measurement of voluntary muscle contractions, which is essential in physical human-robot interactions. Changes in muscle stiffness can be measured with a resonating piezoelectric transducer probe in two ways: frequency shifts and amplitude changes. As the muscle becomes stiffer, the resonance-frequency shift and the signal amplitude change become larger. The frequency shift and amplitude change are mathematically analyzed and simulated. Considering both electrical and mechanical properties of piezoelectric materials, stiffness-based frequency shift and ampli-tude change model are analyzed. As muscle becomes stiffer, the frequency shift increase and the amplitude change decreases proportionally. The sensor was evaluated by comparing the results with those with the force sensor and surface elec-tromyography in the accuracy and response time test from isometric wrist flexion motion. The flexor carpi radialis (FCR) contractions were tested based on maximal voluntary contractions (MVCs). The developed sensor signals are highly correlated with those from reference contraction force sensors. The response time of the aMSS during contraction is faster than that of force sensor, and this means that the aMSS can measure the contraction information before the motion. Moreover, the aMSS can measure muscle contractions through clothing without direct skin contact and despite the reduced signal ratio, this is a distinct advantage over sEMG based sensing. The results show that the aMSS can measure muscle contractions conveniently without direct attachment. The aMSS are tested in long-lasting contraction condition. In this condition, the sEMG signal become larger and the MMG signals decrease due to their physiological properties. The result shows that the aMSS signal change is twice less than that of sEMG and MMG. As a res...