Rethinking Satellite Health Monitoring: AI-Driven On-Board System Sets New Standards, in ESA ARTES programme

Today AIKO, one of the European leader in providing AI-powered solutions for space missions, Tyvak International and IngeniArs proudly announce a groundbreaking achievement in space technology with the successful development of an On-Board Health Monitoring System for satellites.

This innovative project, funded by the European Space Agency's ARTES programme and spanning 18 months, aims to revamp satellite subsystem health monitoring through cutting-edge artificial intelligence techniques. The primary goal of the project is to transform satellite subsystem health monitoring by applying advanced AI techniques for anomaly detection and identification directly onboard, saving time and resources, and improving the operational efficiency of the satellite infrastructure.

AIKO, responsible for the health monitoring software, has collaborated with IngeniARS, the lead in hardware development. A LEO-representative flatsat provided by Tyvak International has played a significant role as a key contributor to the project.

A comprehensive On-Board Health Monitoring System

Developed on AIKO's orbital_OLIVER solution for onboard autonomy, the monitoring system efficiently manages crucial spacecraft resources, including power, bandwidth, telemetry, and payload data. Its highly configurable modular framework ensures adaptability to varying needs. The system works with smart Deep Learning techniques. It gets better even with small amounts of data about unusual situations, making it an effective system that doesn't need much data for training.

Crucially, the system enables the capability to recognize anomalies, as well as to identify their types directly on board, ensuring prompt and appropriate responses. The system was successfully demonstrated leveraging a state-of-art computing architecture developed thanks to IngeniARS' contribution, providing the GPU@SAT ecosystem a fully Space-Qualified and flexible hardware accelerator for Artificial Intelligence, Computer Vision, and High-Performance Computing.

Key Results

The programme achieved the goal of having a high-autonomy onboard solution for monitoring and managing all subsystems, mostly focusing on the Attitude Determination and Control System, demonstrated on actual Tyvak’s FlatSat infrastructure, delivering remarkable results:

• Over 88% of anomalies are correctly detected, improving the accuracy up to 50% compared to a traditional approach
• More than 65% of detections occur in advance compared to traditional approach
• Over 66% of anomalies are correctly classified after detection.

This collaborative effort, marked by seamless coordination among the three engineering teams responsible for software development, hardware development, and infrastructure building and monitoring, represents a significant advancement in ensuring the longevity and reliability of satellites in orbit. AIKO's dedication to advancing space technology is clearly evidenced by the successful integration of innovative AI techniques with state-of-the-art hardware solutions.