E=(AI)² - Edge AI for environmental monitoring from space
Space innovation for onboard data analysis
The E=(AI)² project aims to develop an Edge-AI system for analyzing optical satellite imagery directly on board the satellites. Thanks to this system, data processing occurs via the artificial intelligence integrated into the system itself, right on the satellite, rather than on Earth—as has been standard practice in this field until now. A key component of the system is the self-supervised “feature extractor,” an “intelligent filter” that prepares data for various applications such as cloud recognition, fire detection, and flood monitoring. This approach significantly reduces the time required to detect critical situations.
Agenzia Spaziale Italiana (ASI)
The E=(AI)² project is funded by ASI (Italian Space Agency) under the 2020 "GIORNATE DELLA RICERCA ACCADEMICA SPAZIALE" call, dedicated to developing AI techniques for data processing directly in space. Project partners include Politecnico di Torino, Ithaca, and Argotec.
Edge-AI for environmental emergencies
E=(AI)² utilizes AI technology that analyzes images on board satellites. The goal is to save time and resources by avoiding the transmission of useless or insignificant data to the ground. The system consists of three main modules:
Cloud Screening
Identification of images with excessive cloud cover, which are discarded to avoid unnecessary analysis.
Fires
Rapid recognition of flames and burned areas, essential for timely response.
Floods
Detection of flooding for early warnings and damage prevention.
Speed and precision in sending alerts
The system aims to reduce the time needed to identify events such as fires and floods by transmitting only essential data, thus ensuring rapid intervention. To optimize data transfer to Earth, the system also includes targeted compression of the areas of interest, reducing energy consumption.
Resilient AI architecture
The system architecture is designed to operate in extreme environments, such as space, where radiation can compromise data. Thanks to a modular system and an innovative training method, the project guarantees reliability even under difficult conditions.
The project is made possible through the collaboration between Politecnico di Torino (AI algorithm development), Ithaca (geospatial data analysis support), and Argotec (low-power space hardware).