This works describes an approach to lane-precise localization on current digital maps. A particle filter fuses data from production vehicle sensors, such as GPS, radar, and camera. Performance evaluations on more than 200 km of data show that the proposed algorithm can reliably determine the current lane. Furthermore, a possible architecture for an intuitive route guidance system based on Augmented Reality is proposed together with a lane-change recommendation for unclear situations.

Collision avoidance systems like emergency braking assist systems have demonstrated their effectiveness in increasing the safety of vehicle passengers in various studies. To further increase the effectiveness of collision avoidance systems, the exploitation of the lateral free space by evasive maneuvers is being investigated in this book. This work focuses on methods for integrated trajectory planning and vehicle dynamics control in collision avoidance scenarios by combined evasion and braking.

Die vorliegende Arbeit stellt ein Konzept zur zeitlichen Vorhersage von Fahrstreifenwechselmanövern auf Autobahnen für Systeme zur automatischen Fahrzeugführung vor. Derartige Systeme benötigen ein Verständnis der Fahrumgebung zur konfliktfreien und nachvollziehbaren Durchführung der Fahraufgabe. Dies beinhaltet die Wahrnehmung und Interpretation der Fahrumgebung zur Erkennung und Vorhersage von Fahrmanövern des umgebenden Verkehrs. A concept for time-related forecasts of lane change maneuvers in highway scenarios is presented within the present work. Automated driving systems rely on understanding the driving environment to fulfill their driving task transparently and safely. This involves the perception of the driving environment as well as its interpretation to detect and predict driving maneuvers of road users.

This work proposes novel approaches for object tracking in challenging scenarios like severe occlusion, deteriorated vision and long range multi-object reidentification. All these solutions are only based on image sequence captured by a monocular camera and do not require additional sensors. Experiments on standard benchmarks demonstrate an improved state-of-the-art performance of these approaches. Since all the presented approaches are smartly designed, they can run at a real-time speed.

In dieser Arbeit werden die Grenzen aktueller Odometrieschätzung mithilfe monokularer Kamerasysteme identifiziert und erweitert. Die Basis bildet der Bündelblockausgleich, für welchen Punkte in zeitlichem Bildsequenzen verfolgt werden. Hieraus wird sowohl die Kamerabewegung geschätzt als auch die Umgebungsstruktur rekonstruiert. Mithilfe eines LIDAR wird die Tiefeninformation zu den Kameradaten extrahiert und in das Optimierungsproblem integriert. In this work, the limits of current algorithms for odometry estimation for monocular vision systems are identified and extended. The method is based on bundle adjustment with which tracked feature points in a temporal image sequence are reconstructed and the motion of the camera is estimated. Using an aditional LIDAR sensor, depth information that corresponds to the visual feature points is extracted and integrated in the underlying optimization problem.

Voraussetzung für eine zuverlässige Lokalisierung im Schienenverkehr sind geometrische Karten des Trassennetzes. Diese existieren häufig nicht. Manuelle Verfahren zur Kartierung sind mit hohen Kosten verbunden. Abhilfe schafft die Verwendung fahrzeuginterner Sensoren. Deren automatisierte Verarbeitung erfordert simultan zur Kartierung der Umgebung die Lokalisierung des Fahrzeugs. Abhängigkeiten zwischen der Fahrzeugbewegung und der Karte werden auf diese Weise berücksichtigt.

In this work we present a system to fully automatically create a highly accurate visual feature map from image data aquired from within a moving vehicle. Moreover, a system for high precision self localization is presented. Furthermore, we present a method to automatically learn a visual descriptor. The map relative self localization is centimeter accurate and allows autonomous driving.

Three dimensional modeling and visualization of environments is an increasingly important problem. This work addresses the problem of automatic 3D reconstruction and we present a system for unsupervised reconstruction of textured 3D models in the context of modeling indoor environments. We present solutions to all aspects of the modeling process and an integrated system for the automatic creation of large scale 3D models.

This work is a contribution to understanding multi-object traffic scenes from video sequences. All data is provided by a camera system which is mounted on top of the autonomous driving platform AnnieWAY. The proposed probabilistic generative model reasons jointly about the 3D scene layout as well as the 3D location and orientation of objects in the scene. In particular, the scene topology, geometry as well as traffic activities are inferred from short video sequences.

In this work an approach is presented to model and recognize traffic maneuvers in terms of interactions between different traffic participants on extra urban roads. Results of the recognition concept are presented and evaluated using different sensor setups and its benefit is outlined by an integration into a software framework in the field of Car-to-Car (C2C) communications. Furthermore, recognition results are used in this work to robustly predict vehicle’s trajectories while driving dynamic