DENSE seeks to eliminate one of the most pressing problems of automated driving: the inability of current systems to sense their surroundings under all weather conditions. Severe weather – such as snow, heavy rain or fog – have long been viewed as one of the last remaining technical challenges preventing self-driving cars from being brought to the market. This can only be overcome by developing a fully reliable environment perception technology.
DENSE will lead to the needed key improvements in sensor technology – resulting in a 24/7 all-weather sensor suite – consequently enhancing the safety of automated driving.
DENSE will identify the sensor-based safety requirements for automated driving under varying visibility conditions.
DENSE will specify and develop the necessary sensor technology and validate the resulting fused all-weather sensor suite.
DENSE will – additionally to the technical realisation of an all-weather sensor suite – consider necessary economical aspects to ensure a successful market introduction.
DENSE will start by analysing the contributing factors to car accidents in severe weather. This knowledge will allow for deﬁning the functional and operational requirements needed to build an innovative all-weather sensor suite yielding outline speciﬁcations and system architecture needed for the actual sensor development and integration in cars.
Only a combination of different sensors and the integration of the newest sensor technologies will fulfil the high safety requirements. While conventional radar technology works in bad weather, it does not provide sufficient information to ensure safe autonomous driving.
The project will develop advanced radar with a high angular resolution operating in the 77-81 GHz automotive band, a Short Wave Infrared (SWIR) gated camera sensor with pulsed laser illuminator, and a new LIDAR for improved performance in adverse weather. The sensor suite will also allow for the assessment of road surface conditions.
DENSE will use a smart fusion functionality based on artificial neural networks to fuse all sensor information at pixel level, leading to an enriched and enhanced multi spectral image. For maximizing their efficiency DENSE is planning to implement a high-level fusion platform integration between the single sensors. Ultimately, the system will be integrated in a test vehicle and demonstrated under controlled conditions in a weather chamber as well as under real life conditions, especially autumn and winter conditions, in Central and Northern European.