Autonomous driving in dynamic environments

With the bigger number of cars on our roads every year, also the potential danger to die or get injured in road traffic increases. Most of these accidents are due to human error, while driver assistant systems like ESP help making traffic more safe. To increase this effect, advanced driver assistant systems are under development and a whole body of research is performed all over the world to enable autonomous driving on our roads. In this context in 2004 the work for this thesis was started focusing on the question on how to enable autonomous driving in dynamic traffic scenes, especially considering dynamic obstacles. An autonomous driving demonstrator was built that was a platform for research in the field of autonomous navigation in both structured on road environments and unstructured environments like parking lots. Driving among other traffic participants needs a deep understanding of their behaviors to predict their possible behaviors and act accordingly to ensure safe driving. The algorithms presented in this thesis where widely tested on the SmartTer vehicle at EPFL and ETH and during the preparation of the DARPA Urban Challenge entry of the Tartanracing Team at Carnegie Mellon University in Pittsburgh, Pennsylvania, USA. In the SPARC project (Secure Propulsion using Advanced Redundant Control) the experiences gained from autonomous driving research where also applied to advanced driver assistant systems and showed the strong link between autonomous driving and driver assistant systems.