A FlexibIe Intelligent Environment with Computational Augmentation.
The FINCA represents the integration framework and experimental platform for most of the pattern recognition research activities performed at the Intelligent Systems group. Furthermore, it serves as a platform for scientific cooperations between researchers from different fields as well as between academia and industry. Basically, the FINCA integrates two areas under one roof: a smart conference room and, connected to this, an open and flexible lab-space. Within both areas various sensors, namely cameras, microphones, infra-red sensors etc. are integrated. Electro-mechanical sensors (e.g. light switches) and actors (e.g. light or sunblind control units) are integrated and connected via a KNX installation.
In a smart environment, the goal of computational services is to automatically support the users in an adequate way and to show reasonable reactions to user actions. But which action is appropriate under what circumstances? In order to answer this question, the environment's situational context has to be known. Consequently, the internal state of the environment and the behavior of the users have to be constantly monitored and analyzed. This requires the analysis of large amounts of data from a great variety of potentially distributed sensors. In our research, we seek answers to the following questions: What types of sensors can contribute to the task? How do we realize an infrastructure that is powerful enough to deal with the expected amounts of data and yet remains flexible? And how can humans be perceived as integral parts of a smart environment, not just service consumers, but providing services themselves that could be utilized by the environment?
We developed a simple yet illustrative demonstration of a human centered interaction scheme. Utilizing acoustic sensors in a smart environment we use the obtained information to automatically illuminate the scene accordingly. In our smart conference room, this system continuously reacts to the conversational setting, highlights the active speakers and directs the attention of the audience. The goal of this demonstration is to show, how users intuitively and implicitly interact with the smart environment.
A. Plinge, R. Grzeszick, G. A. Fink. A Bag-of-Features Approach to Acoustic Event Detection IEEE Int. Conf. on Acoustics, Speech and Signal Processing, 2014
Kleine-Cosack, C., Hennecke, M., Vajda, S. and Fink, G. A., Exploiting Acoustic Source Localization for Context Classification in Smart Environments, Ambient Intelligence, LNCS 6439.
Plötz, T., Kleine-Cosack, C. and Fink, G. A., Towards Human Centered Ambient Intelligence, AMI 2008
press review (10/07/2007) (pdf in german)
© Ruhrnachrichten, Westdeutsche Allgemeine Zeitung, Westfälische Rundschau
radio feature on WDR2 (09/07/2007) (mp3 in german)
© WDR 2007.
radio feature on WDR 1live (10/07/2007) (mp3 in german)
© WDR 2007.
radio feature on eldoradio (10/07/2007) (mp3 in german)
© eldoradio 2007.
feature on www.wdr.de (27/07/2007) (webpage in german)
© WDR 2007.