An exact calibration of a vehicle's headlight is mandatory, especially for current and upcoming intelligent headlight systems. Statistics have shown that the headlights of every tenth vehicle in Germany are adjusted too high, greatly threatening night-time road safety through glare for other traffic participants. This paper has proposed a technique for the automatic calibration of an LED lighting system. The method makes use of sensors that are already available in current premium vehicles and needs no active component in the headlight system other than the headlights themselves. It has been demonstrated that the lower light-dark boundary can be found accurately in the images of a nearinfrared camera using sophisticated image processing techniques and combining them with artificial neural networks. The drawback of the method is that a large quantity of example footage with corresponding ground truth data must be made available for optimization of the ANN's parameters. The experiments in this paper were restricted to the usage of a single type of headlight system employing two separate light modules to generate the overall light distribution. Further work must prove that the approach proposed in this paper is also applicable to conventional headlight systems, as currently halogen and xenon headlights exhibit dominating market penetration. Further, the use of a camera sensor of limited quality is appealing. In doing so, improved image preprocessing techniques might compensate for the degradations of the camera sensor. It would then be possible to make use of a camera not exclusively intended for night view applications. Such cameras are more frequently encountered in current premium vehicles, and thus would greatly contribute to function availability.
Headlight range calibration during driving operation
Kalibrierung des Frontlichtbereichs während des Fahrens
2011
15 Seiten, 7 Bilder, 9 Quellen
Conference paper
English
Automatic Headlight for Improving Visibility During Driving
European Patent Office | 2019
|HEADLIGHT RANGE ESTIMATION FOR AUTONOMOUS DRIVING USING DEEP NEURAL NETWORKS
British Library Conference Proceedings | 2019
|Headlight hints for night driving
Engineering Index Backfile | 1928