Machine learning-coupled vertical graphene triboelectric pressure sensors array as artificial tactile receptor for finger action recognition

Wearable pressure sensors have been emerging for recording human biomechanical information during movement and action behavior. When combined with advanced signals processing and analysis methods, these sensors may offer the opportunities of recognizing patterns in specific human actions. In this study, we proposed a machine learning-coupled vertical graphene triboelectric pressure sensors array (ML-vGTEPS array), which could serve as artificial finger tactile receptors for sensing pressure patterns generated by specific finger actions. By utilizing the vertical graphene (vG) with excellent electrical properties and unique nanostructure as the key friction layer, the triboelectric sensor exhibited high sensitivity and broad sensing range. Ten sets of triboelectric sensors are integrated and designed as wearable devices on fingers, which could transmit multi-channel tactile signals under different finger actions with minimal crosstalk. Taking the tracking technical actions of playing table tennis as an example, 16 types of finger tactile signals generated during specific table tennis technical actions were obtained using a self-designed acquisition hardware. Action recognition with high accuracy (98.1%) was achieved by a fully connected neural network (FCNN) deep learning model, serving as validation for the ML-vGTEPS Array in action monitoring and recognition. The exemplary recognition results demonstrated the potential of the ML-vGTEPS Array as a high-performance technique for human-machine interface, intelligent athletic training, telemedicine, and applications in virtual reality/augmented reality.