FingerSonar is a novel sensing technique that can recognize various fine-grained finger gestures by retrieving the acoustic resonance features. It injects acoustic chips (20Hz to 6000Hz) to the body using a surface-mounted speaker from a thumb ring. These chirps are received by four receivers distributed on the wrist and the thumb. By analyzing the frequency responses of these received chips, FingerPing can differentiate up to 22 finger gestures, including touching 12 phalanges using thumb as well as 10 American sign language gestures. A user study with 16 participants showed that our system can recognize these two gestures sets with an accuracy of 93.77% and 95.78% respectively. This project will be appear in CHI 18 conference.
Wearable Tech, Acoustics, Gesture Recognition, HCI, Machine Learning