Recognition of Material Surfaces with Smart Gloves Based on Machine Learning

Nowadays, wearable devices have been extensively studied in the fields of human-robot interaction and motion detection. As a wearable device, smart glove plays a significant role in intellisense. In this paper, a prototype of smart gloves is made by attaching an ultra-thin, sensitive and stretchable ZNS-01 sensor to the surface of ordinary gloves and a data acquisition system is built to collect the data of human hands to touch the surface of the table, wood, plastic, steel, and paper. Time domain characteristics and frequency domain characteristics are extracted as input to train different machine learning models. The prediction results of different machine learning algorithms are compared and the XGBoost shows strong generalization ability and the prediction effect is better. This smart glove can provide a reliable reference for material recognition.