Automation of flight dynamics planning for ESA's XMM-Newton
- Paper ID
88657
- DOI
- author
- company
European Space Agency (ESA-ESOC); Starion Group; Rhea Group; McKinsey & Company; CGI; European Space Agency (ESA)
- country
Germany
- year
2024
- abstract
XMM-Newton, ESA's second cornerstone of the Horizon 2000 Science Programme, launched in 1999, has been pivotal in unraveling cosmic mysteries, from black holes to the origins of the Universe. Despite surpassing its intended 10-year lifespan, XMM-Newton continues to excel, evidenced by its high demand for observation time, with a 6.4 oversubscription factor in 2023. While the ground operations segment has modernized over time, certain processes, particularly those within the flight dynamics team, remain manual. Weekly operational procedures for generating and validating mission scheduling files, incorporating science requests and ground station availability, are among the tasks still reliant on manual effort and expertise. These procedures, although partially automated through legacy scripts and FORTRAN code, necessitate significant human intervention. Engineers meticulously review output logs, diagrams, and file contents to ensure accuracy, given operational constraints such as limited ground station availability and potential equipment degradation, compounded by the mission's extension until 2029 and beyond. Recognizing the need for automation in such repetitive tasks, ESOC's Artificial Intelligence for Automation (A²I) Roadmap identified this use case for further automation through AI. In our collaborative effort with the XMM-Newton flight dynamics team, we developed a solution combining software automation and AI techniques. This solution transforms engineers' roles from task performers to overseers, intervening only as necessary. The AI module optimizes attitude change maneuvers planning, considering operational requirements and leveraging the team's experience to ensure mission safety and integrity. Already operational, this solution positions XMM-Newton as a pioneer in integrating AI into mission operations. In addition to outlining the system's design and technical aspects, we address challenges associated with introducing automation and AI into flight dynamics operations within a constrained environment. Furthermore, we discuss the achieved impact and benefits. The successful integration of AI into a legacy mission underscores its potential to enhance both heritage and future missions. We anticipate that our work will inspire other space missions to leverage AI in overcoming operational challenges.