Φ-lab manages or contributes to several ESA flagship programmes that are expected to massively accelerate the way Earth observation will impact our society
AI for Earth Observation (AI4EO) focuses on harnessing the power of Artificial Intelligence with the vast volume of EO data now available, and as one of Φ-lab’s key initiatives, will exploit the enormous and yet largely untapped potential of AI.
There are many areas of Earth Science, Earth System predictability and Big Data analytics that will benefit from AI, and Deep Learning in particular is pushing the envelope to new levels that are way beyond human performance. The Earth observation research and business communities are now rapidly awakening to these opportunities and implementing this innovation in operation scenarios.
Learn more AI4EO Website
There is an undeniable urgency and moral imperative to act on climate change, the loss of biodiversity and the issue of pollution. Encompassing satellite, climate and societal data we can understand, mitigate and adapt to climate change, to preserve the health of our planet and to better manage natural resources. ESA’s approach to Earth Action extends beyond the space sector, forming partnerships with governments, international and national organisations, Earth science communities, businesses, and citizens to ensure a resilient green future.
Φ-lab explores and supports the integration of disruptive innovation in the Earth Action, mainly ML and Foundation Models, Decision Intelligence, reasoning under uncertainty and innovative computing paradigms such as Hybrid, Quantum and Edge Computing.
Φ-lab contributes to Earth Action by moving away from a data-centric perspective of Earth observation and exploiting all levels of abstraction that the ICT revolution enables, from classical modelling, to forecasting, planning, evaluating up to acting: the ‘Prescriptive AI for Earth Action’.
The Φ-lab Quantum Computing for EO (QC4EO) initiative has been set up to tackle the most computationally demanding EO problems with the much-anticipated Quantum Computing capabilities of the next computer generation. Quantum Computing promises to improve performance, decrease computational costs and solve hard problems in EO by exploiting quantum phenomena such as superposition, entanglement and tunnelling. Use cases examples are mission planning for EO heterogenous constellations, multiple-view geometry integration of EO Images, SAR raw data processing, variational QC4EO image processing, climate adaptation digital twin HPC+QC workflow, uncertainty quantification for remotely sensed datasets. The QC4EO initiative is in collaboration with several research centres and industries, aiming at the creation of a quantum capability in ESA-ESRIN to support programmes such as Destination Earth, Digital Twin of the Earth and Copernicus
Learn more
InCubed is ESA’s Earth observation public-private partnership commercialisation programme that supports the development of innovative and commercially successful products and services. We encourage high-risk/high-reward developments by mitigating the technical and financial risks through co-funding of business ideas.
Learn more
Destination Earth (DestinE) is a European Commission initiative to develop an ambitious AI-driven decision-support system based on a very high accurate digital model of the Earth, to monitor and simulate natural and human activity, along with developing and validating policy scenarios. ESA participates in the initiative by leading the overall effort, in collaboration with EUMETSAT and ECMWF. ESA Member States have initiated a companion programme, the ESA Digital Twin Earth (DTE), to foster the exploitation of DestinE data for Earth observation and innovation.
ESA Member States have initiated a companion programme, the ESA Digital Twin Earth (DTE), to foster the exploitation of DestinE data for Earth observation and innovation.
Φ-lab actively supports the activities of DestinE and ESA DTE to bring the demonstrated analysis and reasoning capabilities of AI technologies to the heart of the operations, along with stimulating further exploitation by various user communities. DestinE will use unprecedented observation and simulation products, powered by Europe’s HPC computers and AI power, to push the limits of computing, forecasting, planning and acting.
The DestinE core service platform (ESA), the data lake (EUMETSAT) and the initial digital twins (ECMWF) will be developed and progressively transferred into operations over the next 7-10 years as part of the Digital Europe Programme. The European Commission (DG-CONECT) coordinates the initiative in close collaboration with other relevant EU programmes or projects like Horizon Europe for research and innovation opportunities, the EuroHPC Joint Undertaking for HPC usage and the European Space Programme for Copernicus data usage.
Learn more
The ESA Φ-lab 3CS initiative is aimed at developing an agile platform for in space testing of the Cognitive Cloud Computing in Space concept which is based on a network of in space agents endowed with sensing (i.e. EO sats) and/or processing and/or communication capabilities where this EO transformative concept can be explored.
This concept is identified as the next step of evolution of EO systems of the future and is an evolution of what is the Φ-sat series to a disruptive concept of in-orbit EO capabilities that exploit all elements of ICT revolution and liberate new business, and operational modes enabled by the availability of in-orbit infrastructure.
Foundation models require large volumes of structured, diverse, and well‑prepared Earth Observation data. Φ-lab supports:
These activities form the entry point of the EO GFM pipeline.
MAJOR-TOMΦ-lab leads the research and development of large EO models capable of understanding multimodal, spatial, spectral, and temporal information at a global scale.
These models aim to:
Our GFMs will evolve toward richer semantic understanding and incorporate predictive capabilities, enabling models to anticipate environmental dynamics and contribute to the development of a dynamic World Model.
With this trajectory, foundation models shift from passive perception systems toward proactive, semantically grounded, and predictive EO intelligence systems.
FAST-EO THOROnce foundation models are trained, large‑scale embedding production transforms Earth Observation data into machine‑readable geospatial representations.
This phase includes:
Embeddings serve as the information backbone enabling semantic EO search, analysis, and automation.
MajorTOM Embeddings on CDSEEO embeddings and foundation model outputs are integrated with advanced reasoning agents and language‑based digital assistants.
This integration enables:
The objective is to bridge human knowledge and expertise with deep, machine-level understanding of EO data, enabling collaborative human–AI systems where domain insight, scientific reasoning, and foundation model representations are tightly coupled.
DA4DTE EVEΦ-lab advances the deployment of EO foundation models onto specialised hardware by focusing on model miniaturisation while preserving generalisation, robustness, and operational reliability.
This includes efforts to:
These compact, yet capable models allow missions to become increasingly autonomous, versatile, and software‑defined, enabling satellites to:
This direction supports the evolution toward AI‑empowered, software‑defined EO missions capable of adapting to new tasks and environmental conditions in real time.
Ф-sat-1Across all these initiatives, Φ-lab supports a full EO foundation model lifecycle:
This holistic framework underpins the vision of EO systems that learn, reason, and operate across the entire space–ground continuum, forming the conceptual basis for Cognitive Cloud‑Computing in Space (3CS) and the emerging Agentic Earth Engine a unified, intelligent layer connecting orbital, edge, and ground segments through adaptive, AI‑driven geospatial reasoning.
