A SOLUTION TO IMPROVE LOCATION DATA ACCURACY FOR MAP-UPDATING FROM MOBILE DEVICES USING THE KALMAN FILTER

Abstract

With the increasing prevalence of mobile devices, it is possible to continuously collect geospatial data directly in the field such as image data and coordinates of objects. However, location data collected from mobile devices often contains noise and large errors, thus direct integration into spatial database poses challenges in terms of accuracy. In the context of the continuous expansion and infrastructure changes of industrial parks in Vietnam, the application of GPS-integrated GIS technology enables automatic and accurate updates of internal road maps within the parks. This article proposes a framework for optimizing point-based data used to update digital maps from mobile devices. The article uses the Kalman filter as a solution to improve the location data accuracy when updating maps from mobile devices. Experimental programming of mobile devices utilizes the Flutter and MapLibre GL libraries, while the desktop program employs QGIS software. The database management systems used are PostgreSQL and PostGIS/GeoServer. Experiments on updating data on real mobile devices demonstrate that the proposed method can effectively limit location errors and minimize the risks associated with anomalous data. By applying the proposed method, the positioning error has been reduced from approximately 8.23 meters to only 3.11 meters. The experimental results also show significant improvements in coordinates accuracy, data deduplication, and map-updating efficiency.