Holding Patterns: Detecting Handedness With A Moving Smartphone


People often switch hands while holding their phones, based on task and context. Ideally, we would be able to detect which hand they are using to hold the device, and use this infor- mation to optimize the interaction. We introduce a method to use built-in orientation sensors to detect which hand is holding a smartphone prior to first interaction. Based on logs of people picking up and unlocking a smartphone in a controlled study, we show that a dynamic-time warping approach trained with user-specific examples achieves 83.6% accuracy for determining which hand is holding the phone, prior to touching the screen.

In IHM 2019, la 31e conf√©rence Francophone sur l’Interaction Homme-Machine (IHM)