James Murphy, Dr. Brian Mac Namee, University College Dublin

The complexity of modern space missions necessitates advanced onboard anomaly detection to ensure their safety and success by identifying unexpected events that could lead to mission-compromising failures.

The challenge is this: to run on-board satellites AI-based anomaly detection needs to work with limited computational resources, stringent power consumption requirements and thermal issues. It must also have the ability to withstand a harsh radiation environment.

Our work has two strands: understanding the best deployment options for edge-AI deployments in space environments and developing novel anomaly detection techniques that match the needs of the space environment. We are addressing these two strands in collaboration with Réaltra Space Engineering, a leader in Irish space engineering.

Our work will lead to anomaly detection techniques that can be deployed on-board satellites to protect them in the harsh environment of space. The impacts of this work will be techniques that enable safer, more effective space missions. We hope that by the end of the project we may be even able to launch our software on an actual mission!