Biosignals

Our human body emits many biosignals due to physical, or biochemical processes. Examples of these "byproducts" of human activity are electroencephalographic signals (brain activity), electromyographic signals (muscle activity), or electrocardiographic signals (cardiac activity). Further biosignals which are investigated at the Cognitive Systems Lab include the human speech and movement.

Biosignal processing is an extremely interesting field of research which is currently becoming more and more prominent. We investigate the potential of capturing and interpreting biosignals for the purpose of building non-intrusive, emphatic systems that sense the user's needs, learn and adapt to a given situation, and present the optimal solutions to its users in everyday live situations.

For several years, we have been doing research on the potential of using EEG (electroencephalographic) measurements for the purpose of detecting a user's cognitive and emotional state. In the future, these results may help to develop interactive systems which interact with the user in a personalized and flexible manner.

Speech is a biosignal itself - it is the most prominent biosignal in human-to-human communication. Recognition of spoken speech has been one of our research topic for many years; for details see our Speech Processing page.

A central topic of our research on biosignal processing is the recognition of silently articulated speech via EMG (Electromyography), i.e. by recording the electrical activity of the articulatory muscles. Such Silent Speech Interfaces enable soundless, but nonetheless natural communication between humans. In the worldwide research on EMG-based Silent Speech Interfaces, the Cognitive Systems Lab plays a leading role. Further details may be found on the Silent Speech Interfaces site.

Motion recognition can be done both visually and by body sensors, i.e. acceleration sensors. In the context of the CRC 588, we investigate motion recognition by camera tracking.

Contact: Michael Wand (michael.wand@kit.edu), Tanja Schultz (tanja.schultz@kit.edu)