WristRotate - A Personalized Motion Gesture Delimiter for Wrist-Worn Devices

Frederic Kerber, Philipp Schardt, Markus Löchtefeld

In: Proceedings of the 14th International Conference on Mobile and Ubiquitous Multimedia. International Conference on Mobile and Ubiquitous Multimedia (MUM-15) November 30-December 2 Linz Austria ISBN 978-1-4503-3605-5/15/11 ACM 2015.


In this paper, we present WristRotate, a personalized motion gesture delimiter that enables separation of non-relevant motion from gesture input. In an extensive data collection, we acquired 435.1 hours of smartwatch acceleration data during everyday usage. We implemented a gesture recognition system based on Dynamic Time Warping to partition a stream of accelerometer readings to identify possible gestures and to classify them accordingly. Through our analysis, we were able to identify a gesture that is (1) uncommon in daily life; (2) quick and easy to execute and (3) easily and reliably detectable. The gesture is executed by simply rotating the lower arm and wrist outwards and back inwards (twice).


WristRotate.pdf (pdf, 874 KB )

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence