December 12, 2022

One year DAKI-FWS - Successful integration of data sources and development of first applications

December 12, 2022

In December 2021, the consortium project DAKI-FWS  (Data- and AI-supported Early Warning System to Stabilize the German Economy) started under the leadership of the AI department  at Fraunhofer HHI. The researchers of the project are working on an AI-supported early warning system to enable a better assessment of crises such as pandemic outbreaks or extreme weather events. This unique system will consist of various heterogeneous data linkages and can be used across industries. The goal is to protect lives, jobs, land, and infrastructure with AI-based forecasts, and to better assess the impact of crises on the economy.

Within the project, the AI department at Fraunhofer HHI is responsible for the development of new intelligent methods. The HHI researchers supplement crisis-specific data with other socially relevant data to enable forecasts that are more accurate. There are numerous synergies with other HHI projects, for example, the ITU/WMO/UNEP Focus Group on "AI for Natural Disaster Management “(FG-AI4NDM), the SAUBER project , and the KLIPS project . DAKI-FWS is part of the "AI Innovation Competition" of the German Federal Ministry of Economics and Climate Protection (BMWK) and is funded with approximately 12 million Euros.

On November 22, 2022, just under a year after the start of the project, the consortium partners of DAKI-FWS met at Fraunhofer HHI to assess the development of the project. They focused on the progress made in the two main areas: pandemic tracking and response, and forecasting of extreme weather events. In both areas, researchers were able to curate a solid foundation of data and use cases. Based on this, the project partners developed initial services. At Fraunhofer HHI, findings from both areas converge to develop the AI-based early warning system.

In the current project phase, HHI researchers led by Dr. rer. nat. Jackie Ma are finalizing the data basis of the AI system; linking weather and mobility data as well as map information to generate application-relevant risk maps. For example, nowcasting methods are developed to train and test the AI with data at 2-minute intervals. In the next step of the project, additional data sources, such as traffic data, will be included to investigate and accurately forecast the impact of extreme weather events on mobility.

In addition to the AI researchers at Fraunhofer HHI, representatives of Zuse Institute Berlin (ZIB), the non-profit organization Data4Life, the German Federal Ministry of Economics and Climate Protection (BMWK), Charité - Universitätsmedizin Berlin, NET CHECK GmbH, Robert Koch Institute (RKI), the German Climate Computing Center (DKRZ), Hasso Plattner Institute (HPI), Justus Liebig University Giessen (JLU), Budelmann Elektronik GmbH, DLR Projektträger, the German Weather Service, Esri Deutschland GmbH, Fraunhofer IAO, Schönborner Armaturen GmbH, VDI/VDE Innovation + Technik GmbH, and Logiball GmbH participated in the status meeting.

The DAKI-FWS project will run for a total of three years. In the first year, Fraunhofer HHI and the consortium partners created the foundation for developing the cross-industry AI-supported early warning system and took the first steps in its application. In the next two years, the researchers will link more data sources and concretize use cases to enable better and faster decision-making for economic experts and governmental actors based on DAKI-FWS.