Artificial Intelligence (AI) has become a powerful tool in today’s data-rich society, enabling us to collect, analyze, and leverage information in unprecedented ways, even within space programs. An exciting example is the integration of AI into satellite navigation, spearheaded by the engineering teams of the European Space Agency’s NAVISP program, collaborating closely with European industry and academia to revolutionize navigation technologies. As a result, an expanding array of prototype services has emerged, each with diverse applications that contribute to enhancing space and Earth weather forecasting, optimizing the performance of self-driving vehicles and marine vessels, and effectively identifying unauthorized drones in sensitive airspace. The combination of AI and space exploration promises to shape the future of navigation and further advance our understanding and utilization of data-driven solutions.

AI as a Tool in Drone Detection

The NAVISP program’s MEDuSA project is tackling the rising issue of intrusive drones by looking skyward. The problem of drones accidentally or intentionally intruding into sporting events, ports, and critical infrastructure is on the rise. A prominent example occurred in December 2018 when Gatwick Airport in the UK had to be closed for three days, leading to the cancellation of numerous flights due to repeated drone sightings near airport runways.

MEDuSA introduces an innovative radar-based approach that can detect drones in all weather conditions and estimate their trajectories. This approach utilizes GNSS signals as the radar signal source of opportunity for sensors to detect drones within the area of interest. It particularly leverages Galileo signals known for their exceptional stability and incorporates the added-value Open Service Navigation Message Authentication service to enhance robustness and protect against spoofing attacks. MEDuSA’s sophisticated algorithms employ ‘forward scattering detection,’ which detects slight signal phase anomalies caused by the passage of drones.

To further enhance the system’s capabilities, Machine Learning (ML) techniques are employed in combination with predictive ‘Kalman filters.’ This ML-driven data analysis allows the derivation of the drone’s onward trajectory, enabling timely alarms and appropriate countermeasures to be deployed, effectively addressing the drone intrusion problem.

Post Image Credit: ESA