The Discovery Campaign on cognitive cloud computing
Below is the list of Discovery programme projects resulting from ideas submitted through the Open Space Innovation Platform (OSIP) Campaign 'Cognitive Cloud Computing in Space'. For each idea, only the prime contractor is mentioned, but the projects will be implemented by larger consortia.
Dual-camera satellite with onboard AI-based decision-making capabilities (OHB Hellas)
Aiming to demonstrate how AI could enable new capabilities in future Earth observation satellite missions. The project targeted both high spatial and temporal resolution imaging with minimal use of space and ground resources. AI was used to fine-point the high-resolution camera towards specific area of interest, after they have been detected in low-resolution images.
STARCORP: Automated and self-improving follow-up verification of detrimental human-activity from LEO (Trillium Technologies)
Enabling the global monitoring of methane emissions using a ‘tip and cue’ system. The project explored using a non-dedicated satellite that rapidly scans Earth such that it ‘revisits’ the same region of ground often (the tip). When the satellite spots a suspected methane source, it tasks another spacecraft equipped with a high-resolution instrument to look at it in more detail (the cue).
D-TACS: On demand data transformations and auto-calibration in orbit (Trillium Technologies)
Implementing onboard, on-demand, fast and accurate machine learning emulators (hardware or software that enables one computer system to behave like another computer system) for atmospheric correction. This would expand the range of onboard use cases, in particular, verification and measurement of subtle variables characteristic of vegetation, fire risk or fuel moisture.
Beyond the Mission Paradigm: Federated Reconfigurable Infrastructure via Edge Devices in Space (FRIENDS) (Mission Control Space Services)
Using federated machine learning to repurpose and reconfigure computer assets for sustainable lunar exploration. Federated learning is a decentralised learning strategy that could enable the different nodes in a network, such as lunar rovers or landers, to cooperate to learn to distinguish different lunar terrains.
PERTEO: PErsistent Real-Time Earth Observation for responsive disaster management (Deimos Engineering and Systems SLU)
Providing global real-time and persistent disaster monitoring services. This study defined and analysed the theoretical PERTEO mission – a mission that would greatly improve services provided to citizens, enable on-demand disaster services, exploit the satellite-as-a-service concept, and employ on-board processing, intelligence and AI applications.
Cognition: distributed data processing system for lunar activities (KP Labs)
Introducing more autonomy into lunar surface exploration. According to a Euroconsult study, 51 missions to the Moon are planned in 2020–2029, and the lunar exploration market will be worth $2.7 billion by 2029. But the Moon is so far away that transmitting data to Earth is a challenge. This project investigated processing data on board the rover and lander, to send only the most valuable data to Earth and increase the autonomy of the rover and lander.
Commercial cloud computing in space (D-Orbit)
Kick-starting future economies through a network of data centres in space, connected via intersatellite links. An early step towards this vision is already operating in orbit; this study took the concept a step further by proposing an open competition for industry in ESA Member States; winners could fly their proposal in space on existing infrastructure.
Blockchain ecosystem for an autonomous consensus mechanism of federated satellite networks (Parametry UG)
Automating cloud infrastructure in space. Bringing a computing system to a higher level of autonomy and awareness comes with the challenge of agreeing on decisions among multiple agents (machines and humans) and detecting needs in resources to plan actions automatically. To achieve this goal, this study investigated some features from blockchain technology and proposed its application for space traffic management and disaster response.
LEO-GEO4GHG: LEO-GEO for GreenHouse Gases (SATLANTIS)
Detecting and quantifying methane leaks in gas facilities in near real-time with high resolution in space and time. The study explored combining data from Earth observation and meteorological satellites with cutting-edge AI to detect and quantify leaks.
Don't try this at home (Planetek Italia)
Demonstrating the benefits of a cognitive cloud computing infrastructure in space to provide commercial services to Earth observation sensors and space-based data providers. Based on a use case set in the maritime domain, the study provided a preliminary design for cognitive computing in space and prepare a technically and commercially viable implementation plan.
Accelerating the technologies needed for effective space traffic management from space. In particular, the project aimed to investigate hybrid edge and cloud computing to perform near real-time localisation of objects in space.
NEU4SST: NEUromorphic processing for Space Surveillance and Tracking (University of Strathclyde)
Demonstrating the potential of neuromorphic computing (inspired by the human brain) and event-based cameras (inspired by the human visual system), for in-space edge-computing detection and tracking of moving targets.