About

PINEBERRY - exPlaINablE, roBust and sEcuRe AI foR demYstifying space operations

The project aims to establish a framework for secure, explainable, reliable, and trustworthy AI for space applications at the European Space Agency (ESA) by:

1. Establishing catalogues and guidelines for security risks and explainability techniques relevant to AI applications in space.
2. Addressing security and explainability issues in selected real-life scenarios of space applications, including satellite telemetry analysis, satellite imagery, and large language models (LLMs).
3. Implementing security and explainability methods in real operational environment.

We believe that PINEBERRY will be an important step toward not only “uncovering the magic” behind AI algorithms (hence building trust in them in downstream tasks) but also in showing that secure AI (SAI) and explainable AI (XAI) techniques can be effectively utilized to improve them, ultimately leading to better algorithms.

Research project done in collaboration between KP Labs, MI2.AI (Warsaw University of Technology), and European Space Operations Center (ESOC); funded by grant no. ESA 4000144194/23/D/BL "Assurance for Space Domain AI Applications".