Abstract
The main goal in the development of driver support systems is to increase safety and comfort in cars and more general in traffic systems. One part of this topic is to warn the driver in case of dangerous situations, caused by obstacles on the street, especially in bad viewing conditions. For this reasons it is necessary to observe the state of this complex system consisting of the car, the road and corresponding objects as extensive as possible and to provide the information to the driver in a suitable manner. Three important components have to been taken under account in such systems: measuring, evaluation and presentation.
Driver support systems should be designed to supplement the perceptional abilities of the driver and to shorten reaction time. A couple of situations is especially appropriate to be assisted this way. Darkness is one of the main reasons for insufficient acquisition of traffic situations by drivers. Statistical inquiries show that high percentage of the fatal accidents occur at night, while only a small part of the whole traffic takes place during night. Rain and fog are reasons for bad viewing conditions as well.
The requirement on the used complex sensor system is very high for this task. Only the combination of several sensors, all of them suited for the acquisition of the vehicle environment in bad viewing conditions, allows to fulfill the needed overall performance of the system. Under these constraints the presented system uses a microwave radar and a far infrared camera. The data fusion process is based on a special Kalman filter approach, which is addressed in this paper.
The paper is based on current activities undertaken within the project EUCLIDE [GRD1-2000 26801] Enhanced human machine interface for on vehicle integrated driving support system supported by the European Commission within the 5th Framework Programme Competitive and sustainable Growth. The project started in 2001 and is coordinated by Fiat Research Centre (CRF).