@City (Automated Cars and Intelligent Traffic in the City) focuses on autonomous driving and intelligent traffic in the city. The goal is not only to support the driver in urban traffic in the best way possible, but also consider the interaction between the car and other (vulnerable) road users. The environment in the city changes rapidly and therefore, conventional methods which are already used in structured and predictable settings like highways, can not be applied to the urban environment.
The goal of @City is the development and the validation of high performing and robust algorithms for the prediction of different scenarios and the representation of the urban environment. Based on the results, new and advanced driver assistance systems for the city are designed, which support the driver with high precision information and take into account the short response times in urban traffic. Additionally, a common understanding of automated driving in the city is developed, which for example defines a set of minimum functionality and safety-relevant aspects.
The subproject “Localization” provides a high definition reference map, which supports the usual sensors (Camera, Radar, LiDAR) necessary to capture the environment, as an additional source of information. This is essential to describe the complex urban infrastructure with sufficient accuracy. Furthermore, the developed algorithms should be able to localize the own car even when the dGPS-System malfunctions or when there are occlusions which can happen frequently in an urban environment. To tackle these special challenges, the subproject aims to develop methods and concepts which are adapted to the traffic in the city. Moreover, plausibility checks are performed considering an inaccurate localization and noisy measurements, such that contradictions can be detected online with some kind of measure of quality. Finally, the old and inaccurate map will be updated using the available sensor data.
At the end of the project, all developed autonomous driving functions are presented in pilot applications, showcasing for example the crossing of a crosswalk, merging into a roundabout or dealing with dynamic obstacles.