Ear-worn reference data collection and annotation for multimodal context-aware hearing instruments.

In this work we present a newly developed ear-worn sensing and annotation device to unobtrusively capture head movements in real life situations. It has been designed in the context of developing multimodal hearing instruments (HIs), but is not limited to this application domain. The ear-worn device captures triaxial acceleration, rate of turn and magnetic field and features a one-button-approach for real-time data annotation through the user. The system runtime is over 5 hours at a sampling rate of 128 Hz. In a user study with 21 participants the device was perceived as comfortable and showed a robust hold at the ear. On the example of head acceleration data we perform unsupervised clustering to demonstrate the benefit of head movements for multimodal HIs. We believe the novel technology will help to push the boundaries of HI technology.

Unsupervised monitoring of sitting behavior.

In recent years, the monitoring of sitting postures was discovered to be a promising measure of healthy sitting behavior, comfort, physical wellness and emotions. Most state-of-the-art systems for monitoring sitting behavior are based on supervised methods that are limited to a fixed set of classes. We present a method that does not rely on training but distinguishes between different postures autonomously. We designed and implemented a system to monitor sitting behavior in an unsupervised manner. Based on the pressure distribution acquired from a pressure mat we generate prototypes of sitting postures. The prototypes are stored in a database and serve as reference for comparing and classifying incoming pressure data. The system relies on only a few, interpretable system parameters and performs in real-time. We conducted an experiment with a collective of 8 subjects and recorded the data of 16 different postures for each subject. Our proposed method generates on average 15.57 prototypes of postures. This reflects well the 16 postures that actually occurred in the experiment. In 91% of all cases an unambiguous assignment of a posture to exactly one generated prototype was achieved. On the other hand an unambiguous assignment of a prototype to a posture was obtained in 86%.