FMG-based body motion registration using piezoelectret sensors.

Abstract

Body motion registration can provide plenty of muscle activity information of human beings, which is applicable in the control of human-computer interfaces or real devices. Forcemyography (FMG) is a method to register real-time body motions by measuring the radially directed force distributions that are generated by muscle contractions. In this work, we recorded FMG maps by using a novel type of sensor, the polymer-based piezoelectrets. With five piezoelectret sensor units attached on the surface of thigh muscles, four basic lower-limb motions, leg-raising, leg-dropping, knee-extension, and knee-flexion, were properly captured on four able-bodied subjects. Motion classification accuracies of 92.9%, 84.8%, and 88.1% were obtained by using different recognition algorithms of KNN, LDA, and ANN, respectively. The pilot experimental results demonstrated the feasibility of FMG recording by using piezoelectret sensors, which may provide an alternative method for body motion registration.