A Wrist-Worn Piezoelectric Sensor Array for Gesture Input
Accurately estimating hand and finger poses for recognizing gestures helps solve many technical challenges, such as controlling prosthetics, robotic manipulation, and computer input, e.g. for virtual reality. A major challenge is accurately identifying gestures under different conditions—such as mobile or low-light environments—without hindering hand function. This paper describes a low-cost wrist-mounted device that uses piezoelectric sensors to estimate finger gestures. The signals that are recorded are vibrations and shape changes that occur at the wrist due to muscle and tendon motion. An array of six piezoelectric sensors was affixed to the inside of an adjustable wrist strap. A user study was completed. To identify when a subject made a finger tap gesture, a touch graphics tablet recorded when a fingertip contacted the tablet surface. Piezoelectric signal features were computed over timing windows coinciding with a gesture. The features were used in the training of a support vector machine classification model. The results indicate the viability of using piezoelectric sensors to classify finger tap gestures, with a mean classification accuracy of 97% for tap gestures made with each of the five fingers.