Machine Learning-Assisted Intelligent Gesture Recognition and VR Space Communication Smart-Glove System Using Strain Sensors

Gesture recognition based on machine-learning approaches plays a pivotal role in advancing Human–machine Interaction (HMI) and robotics. Typical smart glove solutions with flexible sensors can only recognize discrete gestures, not sentences, and fail to meet the daily communication needs of signers. Moreover, current interfaces for displaying recognition outcomes are mostly confined to smartphones or computers, limiting seamless communication between signers and non-signers. Herein, we propose an artificial intelligence hand gesture recognition HMI system consisting of an intelligent sensing glove, a machine learning module, and a virtual reality interface. This system represents a substantial leap forward in gesture recognition technology with adaptability, comfort, and lightweight construction. Enhanced by advanced machine learning algorithms, it achieves remarkable accuracy across a diverse range of gestures—including 10 numeric gestures, 14 alphabetic phrases, and 5 sentence patterns—boasting 99.5% accuracy for numeric and 99.3% accuracy for alphabetic recognition and an impressive 98.8% accuracy for sentence-level interpretation. The system projects recognition results into a virtual space, translating gestures into text in real time. This functionality facilitates not only human–machine interaction but also inter-human communication, particularly bridging the gap between signers and non-signers. It underscores its potential for practical implementation in next-generation HMI systems.