Helmholtz AI
In the last decades, an urgent need for data science has grown across all Helmholtz research domains: more and more data is generated by sensors, observations, experiments, simulations and users. Knowledge extraction from these huge and diverse data sets is impossible with traditional methods. Hence, expertise in and methods from machine learning and artificial intelligence have become decisive prerequisites for future scientific breakthroughs. The Helmholtz association decided to create a structure in which already existing and currently developed AI abilities can be shared, extended, and lifted to a new level through cooperation across domains. One element of this strategy is the Helmholtz Artificial Intelligence Cooperation Unit (HAICU), which consists of one central entity (HAICU Central) organizing communication and collaboration as well as foundational research in the area of data science and a set of local entities (HAICU Local) which are embedded into the various domains yet collaborate across domains through an innovative voucher and project system.
DLR represents the Helmholtz Research Field Aeronautics, Space and Transport (German acronym: LRV) and has a long history of experience in applied AI research. A recent, though preliminary, publication assessment, e.g., has shown that DLR alone publishes more than 25% of the AI-relevant journal papers of the whole Helmholtz association in the past five years.
However, AI expertise is very unevenly distributed in the Research Field LRV. The major contributors to this research area are Earth observation and robotics – with different research foci. A high potential for AI applications in LRV also lies in the topics autonomous driving, traffic research and aeronautics. Many other research groups in LRV, however, have little experience so far, although they could greatly benefit from AI methods. We also see a great need and benefit of DLR-developed AI expertise in other Helmholtz Research Fields.
Earth observation (EO) is a special case, because (i) with the Sentinel satellite fleet of the European Copernicus program and other missions EO has become a big data topic par excellence and (ii) the program topic EO has strong thematic ties to and high cooperation potential with the whole Research Field Earth and Environment.
Motivated by these factors, we propose a HAICU Local unit
“MASTr: HAICU Munich @ Aeronautics, Space and Transport”
It consists of a Young Investigator Group (YIG) in the application domain Earth observation and a High Level Support Team (HLST) providing the expertise from Earth observation, robotics, computer vision and an HPC/HPDA support unit. It involves three DLR institutes:
- Remote Sensing Technology Institute (IMF), Oberpfaffenhofen – coordinator
- Institute of Robotics and Mechatronics (RM), Oberpfaffenhofen
- Facility Simulation and Software Technology (SC), Cologne
The majority of the proposed HAICU Local unit will be located in the Munich area and will have strong interfaces to MuDS, the “Munich Data Science School @ Helmholtz, LMU and TUM” in which DLR is partnering with other Helmholtz centers and two major universities in order to build a sustainable structure for training and recruiting data science experts for research and practice.
The HLST covers a comprehensive package of AI-related services. It is supposed to support both projects within LRV as well as cooperative projects with other Helmholtz centers.