Abstract
A technique for extracting information from user arm gestures that applies pattern recognition to arm muscle biosignals (electromyogram, or EMG), and is able to recognize six basic gestures in real time. The technique analyzes unconstrained free-space movements, and is able make an identification before completion of the gesture. Two electrodes placed on the user's forearm are used to acquire EMG data. Statistical analysis matches incoming data against a training set. The program outputs a gesture by name to a process allowing control of musical playback. We propose a demonstrator whereby a simulation of a personal music player is controlled by hand-arm gestures.

Modules

nWear biosensor hardware

Pattern recognition - Gesture identification

See the video

Biosensors in artistic performance

download PDF
(2.2MB)