DLR – Partner’s Feature Presentation

Hours

DLR – German Aerospace Center

Earth Observation Center (EOC),
German Aerospace Center (DLR)

To the website

The German Aerospace Center is the national aeronautics and space research center of the Federal Republic of Germany. DLR’s research and development work in aeronautics, space, energy, transport, and security is integrated into national and international cooperative ventures. Besides, DLR is the umbrella organization for the nation’s largest project management agency. DLR’s research portfolio ranges from fundamental research to innovative development of products in collaboration with industry partners. DLR provides expert advisory services to the government. Research and development activities at the Remote Sensing Technology Institute target remote sensing technologies. These activities go hand in hand with the German Remote Sensing Data Center’s expertise in large-scale geo-data processing and storage, e.g., through the terrabyte initiative.

What is your organization’s role in the project? What unique contribution does it bring to the team?

For Embed2Scale, the team at EOC’s Earth Observation Data Science department contributes its expertise in self-supervised learning for remote sensing applications. In particular, PI Conrad M Albrecht built an academic track record for multi-modal artificial neural network architectures fusing the Sentinel-1 (radar) and Sentinel-2 (multi-spectral) missions. Lately, Conrad and team researches foundation models for hyperspectral satellite data through DLR’s EnMAP mission launched in 2021, cf. https://www.dlr.de/en/eoc/latest/news/2024/enmap-enters-the-era-of-big-data-for-hyperspectral-foundation-models . Rikard, Michael, and Conrad will lead the research exploring the potential of AI-compression algorithms for federated deep learning. Committed to open science and open-source, in collaboration with the terrabyte initiative, they will contribute to openEO standards that support AI-compressors.

Why is this project important for your organization?

Embed2Scale strengthens DLR’s international collaboration with academia, government organizations, and corporate research for distributed remote sensing analytics with state-of-the-art in artificial intelligence. The project puts to the test the potential of large-scale foundation models in the context of big geospatial data federation to manage petabytes of Earth observation data.

Who are the key people involved?

Conrad M Albrecht

Rikard Vinge

Michael Lukas Marszalek