In the Spotlight

ESA picks two Thales Alenia Space AI projects to test on Φsat-2

In the Spotlight

ESA picks two Thales Alenia Space AI projects to test on Φsat-2

Available in

    The European Space Agency (ESA) has selected the IRMA and PhiFireAI projects, led respectively by Thales Alenia Space’s France and Italy teams, and will test their technologies in orbit on the Φsat-2 microsatellite. This 6U CubeSat planned for launch in June 2024 will demonstrate the contribution of artificial intelligence (AI) on Earth observation satellites to help solve social, economic and environmental challenges.

    A long selection process

    Early in 2023, ESA launched an open challenge to select two innovative AI experiments to fly on the Φsat-2 microsatellite. The competitive selection process, initially involving 75 teams around the world, took over a year and was completed in March 2024. The two winning projects are IRMA (Image analysis for a Responsive Mission with AI) and PhiFireAI from Thales Alenia Space, further consolidating our position in orbital AI.

    PHISAT

    © ESA

    The IRMA project on Φsat-2

    IRMA is an R&D project led by the Saint-Exupéry Technological Research Institute (IRT) in France. Thales Alenia Space and its partners are helping drive this project, which is developing technologies to improve the responsiveness of satellite constellations and meet new needs in the Earth observation market by using AI to convert raw imagery data into interpreted, actionable information.

    The application use case developed by IRMA, which will be tested on the Φsat-2 satellite, involves AI-powered analysis of vast bodies of water to detect potential anomalies directly from orbit, such as oil spills, harmful algal blooms (HABs) and large-scale sediment discharge into seas and oceans.

    MARINE ANOMALY

    This type of monitoring will have two benefits. First, it will enable the most important imagery and data to be prioritized. The images with the highest anomaly scores will be downlinked by the satellite as a priority. Second, it will enable alerts to be raised in real time. When a major event is detected — like an oil spill, for example — an alert will immediately be sent to the relevant authorities, so they can take swift, appropriate action. Conversely, if no anomalies are detected, the images can be deleted directly on the satellite, helping limit the amount of data downlinked to ground stations, optimizing bandwidth use and saving analysis time for system operators.

    The PhiFireAI project

    The solution developed by Thales Alenia Space’s Italy teams is designed to monitor land areas in the event of wildfire. By using AI to process satellite imagery, it will be possible to determine whether each image acquisition in progress indicates the presence of a fire. If a fire is detected, the system will analyze the entire image to locate burnt areas, calculate the size of wider affected areas and detect the presence of water bodies and safe zones.

    The innovation lies in its ability to run the entire algorithm on space-qualified hardware and provide all-important information to end-users in real time.

    The application developed by the PhiFireAI project team will offer huge added value by providing information for current fire monitoring databases. It will also be possible to autonomously detect hazards directly from the satellite. All testing of the model’s performance and adaptability has been conducted in collaboration with Ubotica Technologies (Ireland) and CGI (Italy).

    AI’s growing role in space

    Thales Alenia Space is involved in other projects to develop orbital AI. We’re helping develop a cloud detection technology for the CHIME mission (Copernicus Hyperspectral Imaging Mission for the Environment), one of the six new missions in Europe’s Copernicus program for environmental monitoring.

    Edge computing is another of our core competencies. Working with Microsoft, we’re involved in the IMAGIN-e mission (ISS Mounted Accessible Global Imaging Nod-e) to gather unprecedented Earth observation data from the International Space Station. The goal is to demonstrate the capabilities and operating modes of an architecture designed to allow advanced computing in space, where AI is set to play an increasingly major role.