Gesture-based wearable devices, such as Gesture Pendant or Sixth Sense, leave the user's hands empty. This allows users to abandon interaction with the device instantly when the physical environments requires it, e.g., if the user stumbles. In theory, this gives this class of devices the potential to be used during a mission critical "primary" task, such as driving or rock climbing.

In this paper, we explore this class of devices. In particular, we investigate how to engage in interaction quickly--because such scenarios typically allow users to take their eyes off the primary task only briefly (microinteractions). We present PinchWatch, a one-handed device with a wrist-worn display. Users invoke functions by pinching (i.e., thumb presses against finger or palm); they enter parameters by performing sliding or dialing motions with the thumb on the palm; or they move the hand as a whole (peephole interaction). All user input is tracked by a chest-worn camera observing the user's hand and wrist. Complementing this camera with depth sensing capability allows for tracking additional gestures, for use in sunlight, and for recognizing pinch gestures when the hand is not directly facing the camera.

Our design minimizes interference with the user's primary task by (1) building on large number of single purpose gestures, thereby avoiding modes and menus (2) by offering tactile feedback and eyes-free use for most interactions through pinching, and (3) requiring only a single hand, allowing the other hand to stay on the primary task at all times.