Expanding networks entails expensive civil engineering works – optimized route planning harbors great savings potential here, but also requires reliable data. To this end, Deutsche Telekom AG has recently started leveraging geo-mapping data, which is collected by survey vehicles, via cameras and laser scanners. This 2D and 3D data provides geometric information, as well as details about the ground, the local vegetation, and any street furniture such as advertising pillars, benches, and WCs.

“To be able to plan efficiently, the evaluation of these vast quantities of data needs to be automated,” explains Alexander Reiterer from the Fraunhofer Institute for Physical Measurement Techniques IPM. The researchers have developed software for just that, which automatically recognizes, localizes, and classifies the relevant items in the measurement data. Deep learning is used to analyze the data: The algorithm is ‘trained’ to recognize typical street obstacles on the basis of comprehensive data, and can then add semantic information to the measurement data. Deutsche Telekom AG then receives a digital plan of the area for automated route planning.