TouchSense: Classifying and Measuring the Force of Finger Touches with an Electromyography Armband

Vincent Becker, Pietro Oldrati, Liliana Barrios, Gábor Sörös

Abstract: We present TouchSense, a system to classify and to compute the force of finger touches using an inexpensive, off-the-shelf electromyography (EMG) armband. From EMG input only, we classify the finger touches and estimate the force applied when pressing an object or surface with the thumb, forefinger, or middle finger. We propose a novel neural network architecture for finger classification using EMG data. Our system runs in real time and only utilizes the Thalmic Labs Myo EMG armband and an Android smartphone, thereby being wearable and mobile. We showcase one application for our system, which controls the brightness of a lamp.