The Helmholtz Association is creating a forward-looking network for applied artificial intelligence (AI), the Helmholtz Artificial Intelligence Cooperation Unit (HAICU). HAICU will develop, implement and distribute AI processes, for example for the analysis of complex systems in the areas of climate, energy, transport and health. The German Aerospace Center (DLR) located in Oberpfaffenhofen near Munich has been awarded the contract for the HAICU local unit in the aviation, aerospace and transport research areas. Artificial intelligence and machine learning are powerful tools to develop solutions to these large and complex tasks. DLR is leading the field of applied AI in Earth observation and robotics, but also sees great potential for further research issues such as autonomous driving and aeronautics.
DLR is receiving funding from the Helmholtz Association with which it will establish a junior research group in the field of AI in Earth observation as well as a high-level support team, i.e. an AI pool of experts that will support other institutes and HGF centers in AI projects and implement joint projects with them.
AI in Earth observation
“Globally available geoinformation derived from Earth observation satellite data has become indispensable for many Earth-related scientific, political or planning issues,” says Professor Xiaoxiang Zhu, head of the EO Data Science department at the Remote Sensing Technology Institute and head of DLR’s HAICU unit. Geo- and environmental sciences, sustainable development, resource management, civil security, disaster management and decision support are prominent examples. AI plays an increasingly important role since these methods are far superior to previous algorithms in Earth observation, especially with today’s big data. In particular, high quality geoinformation requirements and a wide variety of applications require Earth observation-specific AI research and innovative methods.
AI in robotics
Safe interactions with humans are necessary for many robotics applications such as in space travel, in disaster management and in nursing. Others, such as planetary exploration, instead require a high degree of autonomy. In all of these cases, AI methods are very helpful in achieving this security, autonomy and robustness. Therefore, the departments of Perception and Cognition and of Cognitive Robotics at the DLR Institute of Robotics and Mechatronics are developing novel machine learning methods in order to allow robots to, for example, detect and grasp objects quickly and robustly, but also to be able to operate error analysis from previously recorded data sets. Other key research aspects of the institute are the persistence of learning methods to respond suitably to changes, as well as autonomous learning, the ability to make decisions that improve the learning process without the need for much human interaction.
Finally, as the third partner in the DLR HAICU unit, the Institute for Simulation and Software Technology contributes its HPC expertise to enable the use of Earth observation and robotics algorithms for large data volumes and high-performance computers.