In the AI-SEE project, a PENTA EURIPIDES² labelled research project is taking seriously the mission impossible – 24/7 automated driving in poor visibility conditions. 21 partners comprising world class players on the OEM (Original Equipment Manufacturer) and supplier level will join forces over a period of three years to build a novel, robust sensor system supported by artificial intelligence for low visibility conditions. The result will be a robust, fault-tolerant multi-sensor perception system. It will be functional in practically all lighting and weather conditions in 24h/365-day mode enabling safe automated driving at SAE Level 4. The project led by Mercedes-Benz AG kicked off on June 2021.
The project is co-developing broad spectrum of complex hardware and software development needed to tackle this currently considered blind spot of automated vehicles. Specifically, the state-of-the-art fault tolerant multi-sensor perception system supported by AI processing scheme will be presented and the related R&I challenges of implementing self-diagnosis, adaptation and robustness discussed. Furthermore, the requirements for inclement weather data collection and sensor near transformation models for synthetic datasets generation will be outlined.
On 31 August and 1 September, 2023, the AI-SEE consortium convened for its 2nd Consortium Meeting at the VTT premises in Tampere, Finland. This gathering brought together over 30 participants with the primary objectives of reviewing the technical accomplishments achieved during the 2nd project year and aligning efforts for the project’s completion in 2024.
During the meeting, notable progress was showcased through the presentation of project demonstration vehicles by VTT, Unikie, Magna/Veoneer, and Patria. This marked a crucial integration milestone in the validation process of the AI-SEE platform.
Additionally, a collaborative testing campaign took place in Padasjoki, Finland, where the Patria off-road land demonstrator, equipped with automotive sensing components, underwent successful testing on forest terrain typical of Finland. This achievement underscored the adaptability and effectiveness of the AI-SEE perception system in diverse domains.
