Dublin based Réaltra Space Systems Engineering, a division of Realtime Technologies Ltd, has unveiled a project funded by the European Space Agency (ESA) under the General Support Technology Program (GSTP) called 'Machine Learning-Based On-Board Autonomy, Failure Prognostics & Detection on Small Satellites'.

Working with subcontractor Ubotica, Réaltra is leading the project which aims to change the current paradigm of Fault Detection Isolation and Recovery (FDIR) and prognostics in satellites. 

Low-latency inferencing capability

The project has recently completed a major milestone, the Preliminary Design Review (PDR). The team at Réaltra aim to offer a dedicated low-power high performance AI/ML acceleration platform which is capable of running AI and ML algorithms with low-latency inferencing capability. This means satellite data will be processed quickly through a trained AI model with respect to anomaly detection and prognosis before appropriate actions are taken.

In this ESA funded 'Machine Learning-Based On-Board Autonomy, Failure Prognostics & Detection on Small Satellites' project, Réaltra are utilising both test campaign and spaceflight data sets from Ireland’s first satellite, EIRSAT-1 (Educational Irish Research Satellite-1), in order to analyse and understand FDIR models and techniques. 

Artificial Intelligence (AI) and Machine Learning (ML) techniques are gaining interest and traction in various domains across the space sector.

The implications of introducing AI/ML driven monitoring systems include the potential to decrease the operating costs and reduce satellite downtime, while enabling the wider use of COTS (Commercial Off the Shelf) electronics to lower the cost of future satellite avionics systems.

By leveraging AI/ML algorithms in this way, satellites will be able to perform a wide array of tasks autonomously, reducing the need for human intervention and significantly lowering operational costs.

From real-time data analysis to automated anomaly detection, AI-driven satellites are set to redefine the boundaries of what is possible in space exploration and communication.

With more than 9,000 active satellites in orbit as of April 2024, some reports have suggested that there could be up to 60,000 active satellites in space by 2030.

Large constellations of satellites in low Earth orbit are the primary drivers of the increase. AI/ML driven monitoring systems offer the ability to detect and respond to anomalies in real-time, minimising downtime and ensuring continuous operation. This proactive approach enhances the reliability and safety of satellites and of satellite networks. 

The aim of the ESA General Support Technology Programme Building Block (GSTP BB) activity is to prepare and develop enabling capabilities for space systems and the associated sub-system. The intended output is to provide industry with a stronger foundation in order to contribute to new space related system opportunities at national or international levels. 

Conor Sheehan, Irish National Delegate to ESA at Enterprise Ireland, said: “We wish to congratulate Réaltra on securing this ESA funding under the GSTP programme. Ireland’s space sector is rapidly evolving, and this project showcases the innovative capabilities of Irish companies on a global stage.

Key player in the international space economy

"By fostering cutting-edge advancements like AI-driven autonomy in satellites, we are strengthening Ireland’s position as a key player in the international space economy. This success reflects the growing agility and ambition of Irish enterprises engaging in transformative space technologies.”

Fintan Buckley, CEO of Ubotica, said: “With the rapid increase in satellites orbiting Earth, the safe and reliable operation of these satellites is becoming increasingly important.

"In addition to enabling real-time on-orbit decision-making, autonomous operations, and streamlining data processing, applying SPACE:AI to address the on-board FDIR challenge is a significant step forward in maximising the operational lifetime of these satellites.

"Solutions like SPACE:AI are pivotal in enhancing communication, situational awareness, and scalability, ensuring the expanding satellite network adapts to future needs.”

Réaltra Space Systems Engineering said: "As the global space economy continues to grow, we recognise the importance of developing and progressing new technologies to meet current and future demands. We are delighted to have secured this ESA project and look forward to working on new, cutting-edge space technology," said Danny Gleeson, CCO of Réaltra.

Project lead, and space software and AI engineer at Réaltra, James Murphy, said: "With the advent of large satellite constellations, there is an increasing need for these satellites to become smarter. On-board AI will soon be a requirement for satellite missions to enable levels of autonomy never before seen.

"Advancing research in AI for Space is an important part of developing the space industry and preparing for the next generation of space flight."