Craft Prospect launches ESA-funded ATLAS project to advance AI-powered validation for autonomous space software

Glasgow-based space engineering company Craft Prospect Ltd (CPL) has kicked off a new project with the European Space Agency (ESA) to develop AI-driven tools for validating real-world space flight software.

The project, AI Test Library for Autonomous Software (ATLAS), will create a suite of advanced multimodal AI capabilities designed to strengthen the verification and validation processes required for next-generation on-board autonomy. These tools will support the assurance of increasingly complex software systems used across modern space missions. 

Craft Prospect is delivering ATLAS in collaboration with leading researchers from the University of Strathclyde and the University of Edinburgh, bringing together specialist expertise in large language models, multimodal AI, and requirements engineering.

Innovating AI-powered software assurance

By leveraging multimodal generative AI to interpret and correlate spacecraft software, specifications, and operational requirements, ATLAS will enable richer understanding and more reliable validation of autonomous software processes.

The resulting test toolkit will be integrated into the software testing lifecycle of Craft Prospect’s Astral Intelligence Toolbox (AITB) — an on-board autonomy framework due to fly next year on the upcoming Craft Prospect-led ESA mission, OPS-SAT VOLT (Versatile Optical Laboratory for Telecommunications).

Words from the partners

Lucy Donnell, Craft Prospect’s Responsive Operations Lead, said of the project,  “Assurance is core to our autonomous software developments. We’re excited to leverage cutting-edge AI to strengthen and evolve our validation processes.”

Arnaud Bourdoux, Flight Software Engineer at ESA and Technical Officer of the ATLAS project said, “Investigating how AI methodologies can be introduced into software validation processes is extremely relevant as these account for a significant part of the flight software development activities, both in terms of cost and schedule. We are looking forward to seeing the results of the ATLAS project applied to the OPS-SAT VOLT mission.”

Dr Johannes J. Norheim, Assistant Professor (Chancellor’s Fellow) in the Department of Design, Manufacturing & Engineering Management at the University of Strathclyde said, “We are excited to leverage our strong research heritage on AI methods for requirements engineering and apply it to support and strengthen software validation processes on a mission flying to space.”

Dr Alessandro Suglia, Lecturer in Embodied Natural Language Processing & Fellow of the Generative AI Lab, School of Informatics at the University of Edinburgh said, “To guarantee safe, reliable software testing, ATLAS is pioneering multimodal generative AI, making it essential to move beyond simple text and master the full, complex information landscape of the aerospace domain.” 

Preparing for next-generation autonomy in orbit

As autonomous space systems grow in complexity, ensuring their safety, robustness, and reliability is a critical challenge. ATLAS aims to demonstrate how AI-powered validation can enhance trust in autonomous behaviours — a foundational step for future space missions deploying intelligent software for navigation, decision-making, and spacecraft operations.

The project is now underway and will continue through 2027.