ESA is an equal opportunity employer, committed to achieving diversity within the workforce and creating an inclusive working environment. For this purpose, they welcome applications from all qualified candidates irrespective of gender, sexual orientation, ethnicity, beliefs, age, disability or other characteristics. Applications from women are encouraged.
Team and mission
Φ-lab, a division of ESA’s Future Systems Department, aims to fully embrace New Space and be the catalyst for radical innovation in Earth observation (EO). In particular, their vision is to become a kind of “hub” connecting a growing ecosystem of Artificial Intelligence (AI) capabilities across Europe. The power of AI for EO is at present largely untapped, and many challenges still need to be tackled at scientific, applications and capability levels to deliver maximum value from open EO data from satellites for our society and economy. See ESA Φ-lab Explore Office Research & Innovation (R&I) strategy, 2020.
In particular, there is growing interest in “hybrid” AI computing, where training is done in the “cloud” while inference is performed at the “edge” in dedicated embedded AI accelerator chips. Such capability provides the foundation for smart connected sensors operating at the ultimate edge, in space. One example of this growing trend of embedded smart sensors is Ф-sat-1, an enhancement of the Federated Satellite Systems (FSSCat) mission launched in September 2020. It is one of the first experiments to demonstrate how AI (powered by a small Myriad chip) can be used for EO, in this case filtering out useless hyperspectral data due to cloud coverage. Such a technology demonstration experiment could stimulate new opportunities for upcoming ESA missions, including CubeSat missions but also possibly Copernicus missions (e.g. CHIME - Copernicus Hyperspectral Imaging Mission for the Environment). The ability to transmit small quantities of user-relevant insight in real time is crucial for the EO sector.
Candidates are encouraged to visit the ESA website.
Field(s) of activity/research for the traineeship
With the rise of hybrid AI computing and AI at the edge, a new set of challenges is coming for the EO community, arising from selection of the relevant applications that could be run at the edge. This concerns for example (1) the definition of adequate Deep Learning architectures, (2) the creation of synthetic data from existing datasets with simulated sensor capabilities, (3) model training and benchmarking, (4) model reduction via pruning and/or knowledge distillation, (5) use of active learning techniques for continuous updating of the deployed machine-learning architecture.
This applied research opportunity would enable tackling some of the above research challenges, and you are thus invited to suggest research activities in the following areas:
- New machine learning schemes and principles (e.g. unsupervised, active, frugal, automated)
- Hybrid computing and edge implementation (e.g. embedded systems, pruning, distillation).
These research activities should be applied to an EO domain you are particularly interested in and be in keeping with Φ-lab strategic research plans.
In addition to your main research activity, you will have the possibility to contribute to the general activities of the Φ-lab involving industry and research centres, and in general support Φ-lab activities in the community related to your expertise.
In particular, you will:
- Perform applied research in one of the above areas, bringing newly-developed AI technologies to upcoming smart space-borne EO sensors
- Contribute to development of an interactive environment for rapid prototyping and testing of new ideas
- Publish results in top-ranked, peer-reviewed publications
- Promote/share your results and tools through new digital tools, including social media and Jupyter Notebook
- Perform or participate in assessments on subjects of strategic interest to ESA.