Latest revision as of 17:47, 11 March 2021

Abstract

The highly automated driving research is one of the most remarkable developments in the automotive world. Still, ordinary people can also realize the milestones facing more and more built-in vehicle assistant services brought by the new car manufacturers. The navigation system is also part of the 'success story' because the maps are currently high-resolution 3D databases, and permanently better visualization supports the drivers. The creation of these map databases relies on cutting-edge field data capture techniques, like aerial surveys. The paper presents a technology based on aerial photographs' evaluation done by human operators. The captured photographs were transformed into distortion-free orthophotos; then, skilled humans evaluated the expected database elements: road and lane segments. The map creation procedure aimed to serve the content fully compatible with the standards of the industry. This evaluation procedure resulted in a road and a lane-level graph, clearly in harmony with the traffic rules. The CAD-drawings of each operator form an excellent starting point to start new artificial intelligence developments.

Full document