A Practical Guide to an Open-Source Map-Matching Approach for Big GPS Data

Abstract

AbstractThis work shows how map-matching helps to minimize errors in GPS data by finding the most probable corresponding points of the recorded waypoints of a trajectory on a road network. We investigate an open-source alternative for map-matching trajectories called Valhalla, which could replace limited and costly commercial map-matching services. Valhalla is an open-source routing engine, which provides different services, such as path-finding, map-matching, and generating maneuvers based on a path. We build a cloud-based big data analytics framework on Amazon Web Services (AWS) platform for map-matching. This well-established framework is scalable and could process millions of trajectories. Using an example GPS dataset, it is demonstrated how Valhalla can be used for map-matching at scale. The dataset consists of about 18 million trips in the year 2019 that have at least one recorded point in a bounding box surrounding Frankfurt am Main. The map-matching results confirm an adequate performance of Valhalla map-matching, show a reduction of errors by distance calculation, and allow for further street-segment-based analysis.