Fabrication of a Capacitive BaTiO3-PDMS Hand Sign Language Sensor and Its Signal Classification Aided by Machine Learning

Abstract

The detection of sign language signals is an important aspect in the life of people with hearing and speaking impairment. It is one important application in the fabrication of capacitive strain sensors. In these kinds of sensors, the change in electrical properties such as capacitance depicts the change in a body’s movement. It is one of the variables along with sensitivity, repeatability, response time, fabrication cost, and ease of fabrication which are the important factors to consider in the fabrication of sensors. This paper presents the fabrication of multi-strain sensors that can be integrated into hand gloves to monitor hand sign language gestures. The designed sensor is composed of barium titanate oxide (BaTiO3) which served as the sensing material and polydimethylsiloxane (PDMS) as the stretchable substrate. The silver electrodes were inkjet printed on top of the prepared BaTiO3-PDMS composite after surface treatment to allow the adhesion of silver ink. The capacitance values of the sensors were continuously measured and recorded using an Inductance-Capacitance-Resistance (LCR) meter while bending the sensors at different bending angles for multiple cycles. The fabricated BaTiO3-PDMS capacitive sensors exhibited a high response of 42.85%, linearity of 0.941, fast response and recovery time of ~1s, and low hysteresis of less than 2%, and good stability over a thousand cyclic test. Its application as a hand sign language sensor was demonstrated to have a unique combination of sensing signals with respect to specific finger flexion angles that correspond to a specific hand sign language of the English alphabet. The interpretation of the sensing signals were aided by machine learning and depicted a classification accuracy of 92% signifying the ability of the sensor to detect a particular letter from the English alphabet.