Robust pedestrian localization in indoor environments with an IMU aided TDoA system

Abstract

Pedestrian localization systems require the knowledge of a user's position for manifold applications in indoor and outdoor environments. For this purpose several methods can be used, such as a Global Navigation Satellite System (GNSS). Since GNSS are not available in indoor environments or deep street canyons other techniques have to be used. This could be a time based or signal strength based radio localization system. In this paper a Time Difference of Arrival (TDoA) system is used and combined with a low cost accelerometer and gyroscope. Since the localization device is free mountable at the user's body, the raw data of the gyroscope must be rotated in a global frame. Therefore the accelerometer is used to compute the rotation angles in relation to the earth's gravity. The data of the accelerometer is also used for a step detection and a step length estimation as well. To fuse the different measurements an Extended Kalman Filter (EKF) is employed. While the system is initialized the orientation of the user is not available. Therefore an initial phase is prepended where a reduced model is used. In this phase the orientation has to develop while the user is moving. The duration of the phase is dynamic, depending on the quality of the TDoA measurements. Once the initial phase is passed the complete model is used. To evaluate the introduced algorithm experimental results in different environments are presented.