Capacitive sensor-based hand gesture recognition in ambient intelligence scenarios

Abstract

Input devices based on arrays of capacitive proximity sensors allow the tracking of a user's hands in three dimensions. They can be hidden behind materials such as wood, wool or plastics without limiting their functionality, making them ideal for application in Ambient Intelligence (AmI) scenarios. Most gesture recognition frameworks are targeted towards classical input devices and interpret two-dimensional data. In this work, we present a concept for adapting classical gesture recognition methods for capacitive input devices by realizing an extension of the feature set to three dimensional input data. This allows more robust gesture recognition for free-space interaction and training specific to capacitive input devices. We have implemented this concept in a prototypical setup and tested the device in various Ambient Intelligence scenarios, ranging from manipulating home appliances to controlling multimedia applications.