Fusion of data from inertial sensors, raster maps and GPS for estimation of pedestrian geographic location in urban terrain

An electronic system and an algorithm for estimating pedestrian geographic location in urban terrain is reported in the paper. Different sources of kinematic and po sitioning data are acquired (i.e.: accelerometer, g yroscope, GPS receiver, raster maps of terrain) and jointly p rocessed by a Monte-Carlo simulation algorithm based on the particle filtering scheme. These data are processed and fused to estimate the most probable geographic al location of the user. A prototype system was designed, built and tested with a view to aiding b lind pedestrians. It was shown in the conducted field trials that the method yields superior results to sole GPS readouts. More over, the estimated location of the user can be effectively s ustained when GPS fixes are not available (e.g. tun nels) .