Abstract
Vehicles are increasingly equipped with driver assistance systems whose
function is based on environment information. On the one hand, these systems aim to make
driving more comfortable, on the other hand to increase safety. This article deals with
environment recognition of such driver assistance systems and shows the implementation of a
baseline system architecture for sensor data fusion of environment sensors developed in
project PRORETA.
In the first part of the article the basic functional elements of an advanced driver assistance
system application are explained (Sensor Signal Processing, Feature Extraction, Data
Association, State Estimation, Classification, Situation Analysis and Control Algorithms) and
their interconnections are analyzed. Based on this information a baseline system architecture
for sensor data fusion of environment sensors is described. The main objectives of this
architecture are independency from the underlying algorithms, testability and the possibility
to evaluate software modules, extensibility regarding applications and sensors, minimisation
of software redundancy and the possibility to distribute and encapsulate software layers over
different hardware devices.
After this theoretical analysis, the implementation of the architecture on a test vehicle of
project PRORETA is described with information about algorithms used in the components of
the architecture. Finally, advantages of the realized sensor data fusion are shown, using test
scenarios from project PRORETA.
Keywords - Sensor Fusion, System Architecture, Environment Sensor, Separation
Theorem, Advanced Driver Assistance System