Posture and Gesture Recognition for Wireless Musical Gloves

Mimu Gloves are wireless, sensor-enabled gloves for creating and performing music. This project will develop advanced posture and gesture recognition algorithms for this glove so that hand postures (e.g. ‘fist’ or ‘one finger point’) and time-varying gestures can be used as a new way to control the composition and performance of music with computers. We will compare a variety of different machine learning techniques applied to the flex sensor data from our glove. We will also vary parameters such as the number/placement of sensors, the number of postures and the amount of training data available for machine learning. Finally, we will explore the recognition of time-varying postures - or gestures. This project will a) extend the capabilities of our hardware glove so that we can develop a fully gestural music making device, with no comparable market competitor; and b) allow us to capture such information about the movement of the hand that we could extend our potential applications beyond music - e.g. video games, virtual reality, medicine and design. The project will be led by mimu Limited and conducted collaboratively with Dr. Tom Mitchell of the University of West England