The collaboration between the University of Trento and the European Space Agency (ESA) is strengthened with a new project destined to revolutionize space missions. The Department of Industrial Engineering of the University of Trento, under the leadership of researcher Paolo Rech, will explore how artificial intelligence can support orbital operations, reducing errors and improving mission efficiency.
Innovations in orbit
For the next three years, the research team will focus on developing neural networks for space use. These automated systems are designed to support astronauts in in-orbit activities, carrying out operations that require precision and speed. They will be able, for example, to support astronauts in extravehicular activities or in the analysis of planetary data directly in orbit.
Artificial neural networks could also reduce the time needed to process data from satellites, probes and rovers. Currently, transmitting data from Earth to Mars can take several minutes, a significant delay that slows field operations and decisions. By automating part of these operations directly on board, AI systems could make missions more autonomous and responsive.
A crucial obstacle in implementing artificial intelligence in space is radiation. High-energy particles, such as ions and protons, can cause serious damage to electronic devices, generating calculation errors that could compromise the entire mission. “Our goal is to test neural network accelerators with particle beams that simulate the space environment,” explains Paolo Rech. “We need to understand how these networks react to radiation and find solutions to mitigate the damage.”
Tests and future prospects
The project received support from ESA through the “Open Idea for Space” program, which promotes innovative ideas for space missions. As David Steenari, ESA engineer and technical officer of the project, points out: “In recent years, interest in the integration of machine learning applications into space systems has grown. However, the radioactive environment poses a significant challenge to the reliability of these systems.”
Rech’s team has already conducted a first experiment in Finland, with promising results. Further tests will follow in Belgium, England and at the Trento Institute for Fundamental and Applied Physics. These experiments aim to collect a wide range of possible anomalies in neural networks to develop preventive solutions.
The integration of artificial intelligence into space missions promises to open new frontiers in the exploration of the cosmos, making operations safer and more efficient. This project represents a step forward towards using advanced technologies that could transform the way we explore and understand the universe.
