The goal of the CarVisionLight research project is to create »thinking« lighting systems that approach or even exceed the human perceptive faculty. The project is funded with 640,000 Euros by the Federal Ministry for Climate Protection, Environment, Energy, Mobility, Innovation and Technology and is part of the ZKW technology programme for sensor integration. This includes the Dragonfly project, which integrates sensors and cameras into headlights and thus enables automated driving functions.
Light is one of the central safety factors in road traffic. Modern vehicle lighting systems support optimal illumination of the road with automatic lighting functions, cameras and sensors. However, there is enormous potential for improvement in the automated recognition of people, animals and objects, especially at night. For example, there is currently a lack of valid data on the distance from which and with what reliability a pedestrian can be detected by intelligent lighting systems at night. For this reason, ZKW and the Vienna University of Technology have carried out comprehensive measurements to improve automated recognition and increase safety during night-time driving.
As part of the CarVisionLight research project, a camera prototype was installed in a test vehicle and thus - in different traffic situations during day and night - data was collected for camera image processing. With the help of artificial intelligence and the monitoring software of emotion3D the information was analysed and patterns were determined. These patterns can be passed on as light algorithms to intelligent camera systems to improve their accuracy. »The camera-based 3D analysis of scenes under the most difficult conditions at night, at high speeds and over long distances is a great challenge for sensor technology, lighting and image processing. Our intelligent image processing solutions can make a significant contribution to more traffic safety and autonomous driving«, explains Dr. Florian Seitner, CEO of emotion3D.
Realizing a Holistic Approach with Partnerships
With the CarVisionLight project, ZKW is pursuing a holistic approach. Therefore, the company regularly cooperates with universities such as the Vienna University of Technology or the Johannes Kepler University or with specialised companies such as emotion3D. »Together with the partners, we are researching image acquisition and processing as well as adaptive illumination. The cornerstone for the next generation of headlamps is the integration of 3D stereoscopy for environment detection with high-resolution headlamp technology. The ultimate goal is maximum safety when driving at night«, explains Schubert.
Two diploma theses by ZKW on the subject of image recognition are currently being commissioned at the Vienna University of Technology. »We are developing new methods of image processing and machine learning that enable the analysis and interpretation of road traffic scenes even under difficult conditions and during night-time driving. The cooperation with ZKW and emotion3D enables students to deal with current topics of autonomous driving in their diploma thesis, which are relevant for both basic research and industrial applications«, explains Ao. Univ. Prof. Dipl.-Ing. Mag. Dr. Margrit Gelautz from the Institute of Visual Computing and Human-Centered Technology at the Vienna University of Technology.