Experimentation
Pilot 1
Monitoring the batteries of Electric Vehicles via worker robots
Pilot 1 aims to monitor and indentify EVs with faulty batteries using Ai-system secured by the cPAID solution.
The approach is based on CAFA Tech RoboCom portal that used for robot management and CAFA Ai-Centre for data processing and analysis. Real time communication between robots and these services is established through secure 5G network. CAFA worker robots will autonomously inspect EVs, asses battery healt and detect potentially dangerous vehicles. The robots transmit data to the CAFA Ai-Centre, where AI/ML algorithms analyse trends and anomalies that may indicate battery faults. Based on these insights, appropriate action can be taken to mitigate risks.
The cPAID solution secures the entire monitoring system by protecting network infrastructure from cyber threats, defending Ai-algorithms against adversarial attacks, and preventing unauthorized access to the CAFA Ai-Centre. It increases communication security between robots and the Ai-Centre while continuously monitoring cybersecurity risks across the entire Ai-driven system. This comprehensive approach enhances resilience against cyber threats, ensuring reliable and secure EV battery health monitoring.
Pilot 2
Monitoring of wild forests with 5G drones and their charging stations
This pilot aims to monitor and detect wildfires using drones and an AI system secured by a cPAID solution.
The approach is based on the CAFA AI Centre, Ground Control and UAV. A high-speed, low-latency 5G network is envisioned for communication among the system elements. The system with computer vision capabilities helps identify and manage wildfires. The drone records pictures and environmental data including plant health, temperature, humidity, and fire indications as it examines its surroundings. Using computer vision techniques, onboard computers detect wildfires like smoke plumes and flames. The drone quickly sends this data to the CAFA AI Centre and CAFA Ground Control Station, where AI/ML algorithms identify trends and anomalies that indicate wildfire threats.
cPAID will provide the security of the 5G network infrastructure, protecting the AI algorithms against adversarial attacks, and safeguarding the CAFA AI Centre from unauthorized access. This approach ensures resilience against adversarial AI attacks, addresses vulnerabilities within these algorithms, strengthens communication security between devices and the CAFA AI Centre, and continuously monitors the cybersecurity risk status of the whole AI system.
Pilot 3
Monitoring of remote AI-assisted medical devices
The University General Hospital of Heraklion (PAGNI) is a leading tertiary care, teaching, and research hospital in Crete, affiliated with the University of Crete’s Faculty of Medicine. It provides comprehensive healthcare services, specializing in advanced medical treatments, emergency care, and specialized surgeries. PAGNI hosts modern diagnostic and therapeutic units, including ICUs, oncology, cardiology, and neurology departments. As a major academic hospital, it supports medical education and clinical research, integrating cutting-edge technology and digital health systems to ensure high-quality patient care and operational excellence. PAGNI primarily acts as a pilot partner in EU projects focused on medical innovation, digital health, and cybersecurity.
Pilot 4
Securing the AI systems of autonomous ships
PyGemini is a platform-agnostic framework designed for developing, building, and testing maritime autonomy software both on-premises and in the cloud. It leverages Generative AI to address gaps when constructing objects (e.g. ships, ports, birds) from sensor data. For the autonomous vessel Milliamphere 2, PyGemini offers a unified solution to streamline the creation and testing of software, from core autonomy functions to simulations. Additionally, it supports user-friendly interfaces for vessel operators and researchers at the shore control lab, ensuring seamless integration while keeping the human in the loop. This is particularly useful for creating relevant training material. In the cPAID project, PyGemini will incorporate cybersecurity features to counter adversarial AI attacks and enhance the resilience of AI-driven autonomy
Pilot 5
Security Training for experts
Nowadays, AI-assisted systems are increasingly integrated into daily operations, bringing both benefits and cybersecurity challenges. Their complexity, multi-tenant nature, and high performance require experts to develop specialized skills. Traditional cybersecurity threats now intertwine with AI-related risks, making advanced training essential.
IQB’s cyber range provides hands-on cybersecurity training through Capture-The-Flag (CTF) challenges and cyber games, covering cryptography, web security, and more. Real-time feedback and teamwork enhance the experience, simulating real-world cybersecurity operations. However, IQB’s cyber range lacks AI-focused challenges, creating a critical gap. Traditional exercises do not fully prepare professionals for AI-specific threats like adversarial inputs and AI-powered attacks. To remain effective, cyber ranges must evolve.
cPAID will bridge this gap by training expert teams from Greece and Norway, including ECSC participants. It will also serve as the AI-security training platform for ECSC 2027, where participants will defend and attack AI-assisted systems. By integrating AI-focused exercises and monitoring real-world scenarios, cPAID strengthens national and European cybersecurity capabilities, ensuring experts are prepared for future cyber threats.
