Advanced driver assistant systems typically rely on input data from multiple sources: radar, infrasound and video cameras. In order to generate highly reliable and trustable decisions, they blend these signals and subject them to strict plausibility checks. A sensor that makes it easier for the algorithms to find the correct decision has been developed by the Fraunhofer Institute for Reliability and Micro Integration (IZM) along with Fiat's Centro Ricerche and STMicroelectronics.
The multifunctional device consists of a video camera, an infrared LED and two light sensors equipped with Fresnel lenses. "Since fog and twilight feature optically identical spectra, it is difficult to distinguish between the two situations", explained group manager Henning Schröder. "For this reason, the infrared LED emits infrared light which is reflected under fog conditions, but not under twilight".
According to Schröder, it is particularly challenging to capture light signals over a wide aperture angle, to bundle it and to guide it to the corners of the camera chip. This is necessary because the center of the chip is reserved for the actual image. In order to facilitate this approach, the researchers have developed light pipes, manufactured in a hot embossing process. The pipes are implemented as small tubes with reflective inner surfaces which can deflect the light by up to 90 degrees. Hitherto, high-end devices for this purpose used fiber optics. However, they tend to break at low bending radii, they are expensive and need to be installed manually which translates into high costs. The light pipes used in Fraunhofer's new sensor make signal transmission more efficient, help to reduce costs and they shrink the size of the device. Another interesting feature: The system is scalable - it can be expanded by additional light pipes, for example to capture sun radiation.
The development is a result of the ADOSE project, partially funded by the EU. The prototype of the sensor currently is tested by Fiat's Centro Ricerche.