To plan its expansion of broadband in Bornheim in the Rhineland, Deutsche Telekom used a measurement vehicle by the name of I.R.I.S. (Integrated Road Information System) with 360 degree cameras, GPS and laser scanners, which moved along the roads of the town and collected data about the surface quality, building development and the environment. A total of around 300 km of roads were recorded, amounting to about 1 terabyte of data. These data were then analyzed by a software package by the name of KNN (Künstliches Neurales Netzwerk [Artificial Neural Network]) developed at the Fraunhofer Institute for Physical Measurement Technology (IPM) . Its deep learning algorithms cover around 30 different categories and recognize, for example, trees, lights, asphalt and cobblestones. In doing so, it even pays attention to the type of sidewalk paving stones and can distinguish between deciduous trees and evergreens. On the basis of this information, the planners at Telekom determine the optimal routes for laying the broadband cables.
Low light conditions and bad weather, in particular, cause problems. The vehicle could not be used in winter as a result. The system is to undergo further improvement, however, and be available to all German Telekom branches from the start of 2020.