Look at the object. Speak naturally. Let attention resolve the ambiguity.
lookNcontrol is a contextual voice-assistant concept based on the idea that looking at an appliance can provide disambiguating context for voice interaction.
The problem
Voice assistants are often ambiguous. If several devices are present, a spoken command may be clear in intention but unclear in target.
Design & implementation
lookNcontrol uses looking direction to identify which appliance a spoken command refers to.
The user can refer to the appliance without saying its name.
What it does
- Voice interaction grounded by visual attention
- Appliance targeting through looking direction
- Smart-home interaction prototype
- Target selection without naming the appliance
