Fusion routine independent implementation of advanced driver assistance systems with polygonal environment models

Future driver assistance systems will rely heavily on fusion systems and an exact knowledge of the static environment. Several aspects like driving path detection, parking assistants, emergency braking and trajectory planning need the knowledge about non-moving objects in the surroundings of the vehicle. In order to build robust maps of the environment, several possibly conflicting sensor readings must be combined. However, an increased number of sensors and an increased number of driver assistance functions lead to difficulties on integration in the fusion system. Different interfaces must be supported, since sensors provide their data in special formats and functions require each other aspects of the environment. This difficulty has been relaxed by introducing the Fences-architecture and homogeneous interfaces. The architecture leads to several useful properties. This contribution deals with a particular property, the independence of the fusion subsystem from all other subsystems. This property states, that driver assistance functions are executable without modifications, in case the mapping technique is changed or even removed. The property is demonstrated by implementing two different mapping strategies. The environment models created by these strategies as well as an environment model created by a laser scanner are afterwards used to execute a road boundary extractor, which is readily introduced.