The work is a first step towards relaying real time information from space to disaster response teams.
The Oxford team has developed a machine learning / artificial intelligence model called ‘Worldfloods’ designed specifically for deployment in specialized hardware in space on low-cost satellites in Low Earth Orbit.
The model is a flood segmentation model that has the purpose of detecting flood events and significantly improving disaster response operations. It has major implications in bringing down the cost of such technologies and making it accessible for low income countries.
Atılım Güneş Baydin, based at the Departments of Engineering Science and Computer Science, Oxford, said: ‘This will be the first time a machine learning model for this type of task will be actually deployed in space. It's a very significant step in bringing machine learning and artificial intelligence operations to space.’
The international team of eight people conducted the research as part of the ‘Frontier Development Lab (FDL) Europe’, a partnership between the University of Oxford, European Space Agency’s Φ-lab (ESA ESRIN), Trillium Technologies and leaders in commercial AI, such as Google Cloud and Intel.
The team includes Gonzalo Mateo-Garcia, Joshua Veitch-Michaelis, Lewis Smith, Silviu Oprea, Guy Schumann, Yarin Gal, Atılım Güneş Baydin, and Dietmar Backes. Three members are Oxford researchers: Atılım Güneş Baydin (Engineering Science and Computer Science), Yarin Gal (Computer Science), and Lewis Smith (DPhil student at Computer Science).
The two-month FDL programme was hosted at the ESA ESRIN Φ-lab (Italy) and at Kellogg College in the summer of 2019. The team has kept working on the project and kept developing the system until the present day.
It has had input from UNICEF due to the humanitarian aspects of the project.
An article recently published in Scientific Reports covers the details of the work: https://www.nature.com/articles/s41598-021-86650-z