Evaluation of Simple Microphone-based Mechanomyography (MMG) Probe Sets for Hand Stiffness Classification

Abstract

We describe simple to build mechanomyography sensors, with one or two channels, based on electret microphones. We evaluate their applica�on as a source of infor-ma�on about the operator�s hand s��ness, which can be used for changing a robot�s gripper s��ness during tele-opera�on. We explain a data ac�uisi�on procedure for further employment of a machine-learning. Finally, we present the results of three experiments and various machine learning algorithms. �upport vector classi�ca�on, random forests, and neural-network architectures (fullyconnected ar��cial neural networks, recurrent, convolu-�onal� were compared in two experiments. In �rst and second, two probes were used with a single par�cipant, with probes displaced during learning and tes�ng to evaluate the in�uence of probe placement on classi�ca�on. In the third experiment, a dataset was collected using two probes and seven par�cipants. �s a result of the singleprobe tests, we achieved a (binary� classi�ca�on accuracy of ���. �uring the mul�-probe tests, large cross-par�cipant di�erences in classi�ca�on accuracy were noted, even when normali�ing per-par�cipant.

Novel near E-Field Topography Sensor for Human–Machine Interfacing in Robotic Applications

“…Mechanomyography (MMG) approaches typically use microphone- or other membrane-type sensors attached to the user to measure the muscle activation more directly [ 15 , 16 ]. These sensors can reduce the amount of contact between the sensing element and the surface of the skin, requiring only a transmission medium or seal for operation [ 17 ]. These sensors can generate high-quality data with similar noise susceptibility and signal characteristics as EMG data.…”