Newsletter #3 | December 2021
Dear Φ-lab community,
From 11 to 15 September, the biggest ever edition of Φ-week took place as a hybrid event with an unprecedented number of registrations – almost 4,000 – and an amazing variety of topics. The event was characterised by the remarkable diversity of speakers, who came from various technical backgrounds, countries, institutions and professional levels. During the week, new trends in space technologies, future missions and the commercialisation of space activities were addressed, and there was also several significant announcements, including the launch of the InCubed co-funded DFKI AI4EO Solution Factory, and the Φ-lab Investing in Industrial Innovation 2022 event. The latter will be held by our very own Φ-lab Invest Office – more details to come.
The Φ-week agenda included sessions on AI4EO (Artificial Intelligence for Earth Observation) and QC4EO (Quantum Computing for Earth Observation). These discussions attracted particular attention and sparked a lot of interest among the over 450 attendees, with many of the presentations addressing issues linked to climate change and real-life challenges. The topic of Quantum Computing in particular is experiencing increasing popularity, due to its promising capabilities for reducing computational complexity problems. Its application to the field of EO however remains a major challenge, in particular regarding the encoding of Big Data on computers with only moderate levels of qubits.
The other major news item from Φ-week was the ESA Director General's call for ideas on Cognitive Cloud Computing in Space, which was developed within Φ-lab. The call covers various space domains, from EO to telecom and deep exploration, and the top ten proposals will receive funding support for development studies. Ideas should aim to increase the flexibility, autonomy, speed and intelligence of space infrastructures by processing satellite data in orbit to extract actionable information. To this end, the campaign encourages submissions based on Artificial Intelligence and associated fields such as Neuromorphic Computing, particularly as the use of these paradigms for space applications is still under exploration. Many applicants have already responded to the call, and their ideas are currently being assessed.
Continuing with the AI theme, mid-November saw the second edition of the ESA-ECMWF workshop on Machine Learning for Earth System Observation and Prediction (ML4ESOP). With over 1,100 participants from 85 countries around the world, this four-day event included 33 talks from ML4ESOP experts across four thematic areas. The last day consisted of discussions on current ML applications, ML advantages over traditional methods, limitations, challenges and future directions. This year’s edition delivered strong messages of replicable, scalable, and sustainable ML methods, and was a great showcase for the importance of building Earth models and digital replicas of our Earth system, as demonstrated by the ambitious Destination Earth EU initiative.
The Φ-lab team
