Device-free hand gesture recognition exploiting machine learning applied to RFID

Gesture recognition is a recently emerged technology that aims to redesign the interaction between humans and machines. Existing solutions usually recognize the gestures based on camera vision, wearable sensors or specialized signals (e.g., WiFi, acoustic and visible light), but they cope with the drawbacks of high energy consumption or reduced SNR with respect to the surrounding environment, which prevents them from accurately sensing finger movements. In this work, we propose a device-free gesture recognition system that use machine learning techniques on RSSI and phase values from the backscattered signals of an array of tags placed on a plastic plate to recognize different hand gestures. The system has been tested with RFC and SVM algorithms, providing excellent results both in case of stationary hand that in case of moving hand.