Repairing space software with AI

For a spacecraft to carry out its role effectively, not only must all the physical parts, nuts and bolts work well, but so must the onboard software. For this reason, software designed to be sent into space is continually checked from the very beginning of its development.

Ensuring the quality of software inside a spacecraft takes a lot of time and effort. To make the process more efficient, ESA Discovery is supporting a team from Thales Alenia Space to repair software using artificial intelligence following one of the early verification techniques, a process called 'static analysis'.

Arnaud Bourdoux, ESA engineer overseeing the project explains: "Static analysis verification is a part of the flight software development process during which various tools are used to check embedded code against common programming mistakes, quality rules, and dozens of other criteria. Such tools are now a standard part of software development environments, but one of their drawbacks is that their output can be tedious to process."

"The use of AI in this domain may help accelerate the processing of the output from these tools; the AI should be able to suggest corrections for identified problems, as well as recognise the false positives that do not require any modification."

Miguel Fernández, flight software manager at Thales Alenia Space concludes: "Using artificial intelligence will help us develop software for space with increased quality and decreased development time. ESA opened my mind thinking about how to develop software for space more efficiently. Now, with ESA's support, this will lead to new opportunities for onboard software development."