Application of inertial sensors and raster maps for correcting GPS readouts in urban terrain

An electronic system for correcting GPS readouts is reported in the paper. Different sources of data are acquired, i.e.: accelerometer, gyroscope, GPS receiver, raster maps of terrain, that are processed by a Monte-Carlo simulation algorithm based on particle filtering. These data are fused to estimate the most probable geographical location of the system user. A prototype was designed, build and tested with a view of an application for aiding blind pedestrians. The method yields superior results to sole GPS readouts. Moreover, the estimated location of the user can be effectively sustained when GPS fixes are not available (e.g. tunnels).