Real-time computer control using pattern recognition of the electromyogram

A real-time system is discussed which recognizes users gestures using neural networks for pattern recognition of the electromyogram (EMG). A novel application for dynamic computer control is presented. In addition to gesture recognition, an estimate of muscular exertion is provided to facilitate control of continuously varying parameters. The use of this system for a real-time graphical user interface (GUI) is discussed. In this example the direction of motion of the users arm is detected and used to control a graphical slider, which moves in the direction of the users arm. The speed of the slider is proportional to the muscular exertion. The system described has applications to physically disabled users, and can be adapted to an individuals abilities.