ABSTRACT
The article deals with the application of inertial sensors (accelerometers and gyroscopes built in smartphones) for pedestrian positioning in the indoor environment. The problem of using inertial sensors results from their functional principle, which is based on the integration of inertial measurements. A secondary product of integration is the rapid accumulation of relatively small errors of inertial sensors in the actual position and orientation of pedestrian. To eliminate the systematic errors of the accelerometers, model of data processing uses adaptive estimate of the step length. This algorithm uses the actual amplitude of acceleration and the walking frequency to estimate the step length.
The often-applied approach to eliminate the systematic errors of gyroscopes is the approximation of direction of pedestrian motion in four main directions, but this solution assumes the rectangular geometry of buildings. In order to eliminate this limitation, the article proposes new model which is extended by algorithm for identification rooms and a combined calculation of orientation to eliminate the systematic errors of gyroscopes. The article describes the algorithm for identifying the room where the pedestrian is located. On the base of the identified room, the model identifies the room geometry (from floor map of the room) and the corresponding solution (with or without approximation of orientation). In the case of room with atypical geometry, the orientation calculated without approximation and the map matching algorithm is used for verification and correction of the calculated position of pedestrian based on map information of the selected room.
