Crowdsourcing for live maps

GPS-tracks from volunteers can be used to improve maps. In the future, maps may even go live: displaying what’s taking place at the very moment. Master student Karl van der Winden pioneered the auto-updating of geographical information.

Van der Winden is enthusiastic about OpenStreetMap, the Wikipedia of mapmakers. But he also noticed that many of the secondary tags with additional information on roads are often left blank. He wondered if attributes like maximum speed, one way circulation, access and number of lanes could be derived from gps-tracks from users with smartphones.

Such information could contribute in a major way to OSM’s information richness since currently only a quarter of the highways is tagged with maximum speed and only one out of seven one-way streets is properly marked.

Van der Winden discovered that gps-tracks from volunteers (VGI or volunteered geographic information) needed a lot of preprocessing before attributes could be extracted from the data. This way information about one-way traffic, maximum speed, average speed and congestion can be derived.

The number of lanes however proved to be near to impossible to derive from gps-tracks. Van der Winden proposes to use additional data like satellite images or images made in cars to update lane information.

Updating OpenStreetMap with the automatically derived data proved harder than expected. Van der Winden found out that automatic updates were blocked, even though automatic updates, in his view, could be very beneficial for the completeness and the level of detail for OSM in the future.

He mentions navigation systems that would operate better with the proper information on accessibility (different for cars and bicycles), one-way traffic and maximum speed. And also traffic analysts would be helped with adequate data.
Assuming the OSM-organisation will give the resistance to automatic updates, Van der Winden imagines the possibility of live maps: ‘Crowdsourcing might be helpful to acquire a vast amount of data from people. This could lead to live maps which would change almost real-time with events on the road.’ 

Van der Winden (24) is one of the first two master student from geomatics to graduate. This master programme was added to the faculty of Architecture two years ago.

Karl van de Winden, Automatically deriving and updating attribute road data from movement trajectories, thesis supervisor F. Biljecki MSc., 26 June 2014.