Dominik Gusenbauer

Self-contained indoor positioning on off-the-shelf mobile devices

We introduce a novel self-contained seamless positioning solution for indoor and outdoor environments, well suited for and designed to be operated on off-the-shelf mobile phones. Position information is deduced from a combination of GNSS where available, combined with Pedestrian Dead Reckoning (PDR) utilizing inertial measurements and context-aware activity based map matching. The proposed system heavily exploits different types of human movement, such as walking, running, ascending or descending stairs, to improve the employed positioning model. In remaining independent from any external infrastructure, accurate localization is also possible in environments, where the installation and maintenance of such infrastructure does not make sense or is simply not affordable - as for example in a parking garage to guide a user to the next exit or back to his car. Concerning this particular use case we have also implemented an interface for the synchronization of location information between the mobile positioning solution and the car.

Camera based step detection on mobile phones

This paper presents an extension to an indoor positioning system leveraging the camera which is built into current mobile phones to detect steps. Current indoor positioning systems on mobile phones based on pedestrian dead reckoning (PDR) often rely on step length and step frequency estimation. Usually the accelerometer is used to determine these values. Depending on the attitude of the device or the smoothness of the movement sometimes steps are not correctly detected. We propose to use the camera as additional sensor element to detect the forward section of both feet appear and disappear in the camera image and detect the steps based on this information. A comparison to an accelerometer based approach is provided.

Managing Context on a Sensor Enabled Mobile Device - The mSense Approach

We present a novel context acquisition and management middleware for mobile phones based on Qt (C++ based cross-plattform API) for S60 called mSense. In our middleware, sensor nodes encapsulate platform level APIs for seamlessly accessing hardware sensors, simulated sensors or web services. Aggregators (channels) combine information from other nodes to generate new knowledge. For example, GPS, accelerometer and compass could be combined to provide a more accurate inertial sensor node. The output is supplied via a native interface or broadcasted through web services.