SkinGest: artificial skin for gesture recognition via filmy stretchable strain sensors

Abstract

Stretchable sensors are promising in the field of wearable robotics. To date, it is still a challenge to design an artificial skin with thin and sensitive stretchable sensors. In this paper, we present a new artificial skin, SkinGest, integrating filmy stretchable strain sensors and machine learning algorithms for gesture recognition of human hands. The presented sensor has a sandwich structure consisting of two elastomer layers on the outside and one soft electrode layer in the middle. Based on the improved fabrication process, we make the sensor’s thickness down to 150 µm, while keeping the gauge factor (GF) up to 8. Then, we integrate the machine learning algorithms (using LDA, KNN and SVM classifiers) with the stretchable sensors in our SkinGest system for gesture recognition. Supported by the experimental data from different subjects, our SkinGest system succeeds in identifying American sign language 0–9 with an average accuracy of 98%. The results demonstrate that the proposed SkinGest system provides a promising platform for future potential virtual reality and sign language recognition applications.