Abstract
Keywords Driver Assistance System, Lane Departure Warning, Attentiveness Control, Image Processing, Safety
Abstract The most frequent causes of traffic accidents are inattentiveness and misinterpretation of the traffic situation. Both causes are connected with driver information and the way and the speed the driver digests the information. The earlier the traffic participant can realize precisely a change in the traffic situation and he can react to the change, the easier he can avoid the accident or reduce the accident aftermath.
In this regard, the importance of the safe lateral guidance has to be stressed because lane departure almost always results in severe injuries or fatalities of the vehicle occupants. Up to now, the driver has to cope with the lateral guidance himself since there is no assistance system available for this driving task.
In order to develop a lane departure warning system, several sensor systems were tested in an experimental vehicle; infrared sensors in the outside rearview mirrors as well as a mono- and a stereo-camera system. As a matter of principle, camera systems are superior due to the possibility of lane-prediction.
The camera signals are processed by an embedded PC with image processing software. The main algorithm receives the calculated lane and object data over a separate, private CAN-bus.
The basic structure of the main algorithm can be subdivided into four parts. In the first part, all necessary data from the vehicle and the image processing is collected via CAN-bus. The core of the algorithm (part II) calculates the warning state and the volume of the warning basically from the determined time-to-line-crossing (TLC). To avoid false alarms due to a failure of the image processing or recognized active driving by the driver (i.e. braking, overtaking, lane change without turn indicator), a subsequent warning suppression system resets the warning state. If there is no active warning suppression, the warning will be emitted by a sound-generator (part IV) and the in-car speakers.