Mansimran Singh Undergraduate Dissertation 2015/16

Gesture recognition using Electromyography

Supervised by M.Stannett

Abstract

Human interaction with machines has moved from physical analogue controls to touch screens, to provide a more seamless way of action and feedback. This project looks at human machine at a step further: interaction using gesture control, specifically EMG based gesture recognition. 

The project aimed to develop a system for gesture recognition using a Myo, to enable users to play a trivial game of Rock Paper Scissors Lizard Spock against the computer.

A static gesture classifier for streaming EMG data, using windowing techniques and k-nearest neighbour algorithm was implemented and a simple game interface was created: enabling users to play the game against a computer player.