UofG engineers lend support to new ESA-funded project
Published: 11 June 2025
Experts from the Space and Exploration Technology Group are supporting Craft Prospect Ltd on a new contract from the European Space Agency to prototype and demonstrate machine learning algorithms for motion planning in orbital space missions.
Glasgow-based space engineering practice Craft Prospect Ltd (CPL) has won a contract from the European Space Agency (ESA) to prototype and demonstrate machine learning algorithms for motion planning in orbital space missions.
CPL is supported in the project delivery by an expert team from the University of Glasgow’s Space and Exploration Technology Group.
The project, ‘Robust real-time constrained optimal control using machine learning’ (ROC-ML), will examine the challenges in generating safe, optimal paths for navigation in spacecraft with thruster control, covering missions include active debris removal, in-orbit servicing and space station assembly.
The use of AI provides a vital capability in responding to challenging mission scenarios, where targets can be spinning uncontrollably and environments can be littered with ancillary debris and other obstacles.
Now in its preliminary design phase after a successful System Requirements Review, ROC-ML is employing a rapid prototyping, agile approach to product development.
The preliminary design work will be performed in parallel with hands-on development and implementation of the AI-augmented guidance navigation and control (GNC) system. A hardware-in-the-loop simulator is being developed to train and demonstrate the GNC software, focussing on advanced ML paradigms like reinforcement learning and imitation learning.
The project will proceed through its critical design phase into 2026 and conclude next year with a test campaign to validate and demonstrate the AI-optimised motion planning capabilities of the system.
Valentin Preda, project Technical Officer and GNC Engineer at ESA, said of the project: “Autonomous motion planning and control in complex, dynamic, and uncertain environments is a cornerstone technology for the next generation of space missions. This collaboration with Craft Prospect provides an exciting opportunity to explore how advanced machine learning techniques can enhance guidance, navigation, and control systems supporting greater autonomy and responsiveness, and extending the capabilities of spacecraft in real-time decision-making and adaptation.”
Murray Ireland, Head of Engineering at Craft Prospect, said: “This is a fantastic project to be working on with ESA. State-of-the-art and next-generation RPO and ISAM spacecraft face significant challenges in conducting successful, safe and sustainable missions. Leveraging our existing expertise and solutions in on-board AI and the University of Glasgow’s knowledge in next-gen satellite GNC approaches, we’re excited to show how AI can be employed with high impact and minimal risk.”
Dr Kevin Worrall and Dr Matteo Ceriotti, of the University of Glasgow’s James Watt School of Engineering, said: “We’re excited to work with our friends and neighbours at Craft Prospect and ESA on this project. We have been pioneering research on machine-learning-based attitude control algorithms, and this project takes this research one step forward, focusing on the problem of the robustness of the machine-learning-generated solution, to guarantee safe and reliable solutions as required in many of the challenging and autonomous space applications.”
First published: 11 June 2025