Automatic update of road attributes by mining GPS tracks

Automatic update of road attributes by mining GPS tracks

van Winden, Karl
Biljecki, F. (TU Delft Urban Data Science)
van der Spek, S.C. (TU Delft OLD Urban Design)

2016

Despite advances in cartography, mapping is still a costly process which involves a substantial amount of manual work. This article presents a method for automatically deriving road attributes by analyzing and mining movement trajectories (e.g. GPS tracks). We have investigated the automatic extraction of eight road attributes: directionality, speed limit, number of lanes, access, average speed, congestion, importance, and geometric offset; and we have developed a supervised classification method (decision tree) to infer them. The extraction of most of these attributes has not been investigated previously. We have implemented our method in a software prototype and we automatically update the OpenStreetMap (OSM) dataset of the Netherlands, increasing its level of completeness. The validation of the classification shows variable levels of accuracy, e.g. whether a road is a one- or a two-way road is classified with an accuracy of 99%, and the accuracy for the speed limit is 69%. When taking into account speed limits that are one step away (e.g. 60 km/h instead of the classified 50 km/h) the classification increases to 95%, which might be acceptable in some use-cases. We mitigate this with a hierarchical code list of attributes.

http://resolver.tudelft.nl/uuid:e54e0ce3-5118-423a-9800-885a5ffa04d8

https://doi.org/10.1111/tgis.12186

1361-1682

Transactions in GIS, 20 (5), 664-683

Institutional Repository

journal article

© 2016 Karl van Winden, F. Biljecki, S.C. van der Spek