: Siegle, G.; Geisler, J.; Laubenstein, F.; Nagel, H.-H.; Struck, G.

Abstract
In order to drive a vehicle without manual intervention, short range information about the visible surrounding in front of the vehicle is extracted by machine vision: a fast, region-based tracking of lane boundaries and - at a lower frequency - an evaluation of the entire video image in order to detect intersections. This is combined with internal map-based knowledge about the road network, which is supplied by the navigation system TravelpilothighR commercially offered by Robert Bosch GmbH. The modules have been integrated into an experimental vehicle. Given a desired goal-position on a suitably selected road network, the vehicle is able to perform the necessary driving mission autonomously.