Discover how AI-powered synthetic humans are advancing mission readiness by accelerating operator decision-making, enabling high-fidelity judgmental training, refining mission-critical communication, and adapting to complex, high-consequence environments.

This presentation focuses on the application of technologically advanced solutions that, with minor adaptations, can be successfully utilized across various sectors and industries, including defense.

We will primarily explore the use of artificial intelligence for anomaly detection and predictive maintenance in manufacturing, forecasting sea tides and river flows, and generating synthetic data. What can we learn from other industries, and how can we transfer successfully implemented innovations into the defense sector?

AI enhances the value that OPTIMA — our multi-agent simulation environment for data farming and analytics of adversarial scenarios on the operational level — delivers to our clients. We share our experience in applying AI across three key areas: modeling agent behaviors, analyzing simulation outputs, and informing agent decision-making.

We will present the applicability of geospatial artificial intelligence for the automatic detection and recognition of military and civilian personnel, vehicles, and objects, as well as the importance of integrating decentralized AI systems that effectively support decision-making. Additionally, we will introduce a completed research and innovation project by the Ministry of Defence of the Republic of Slovenia, which utilizes sensors (such as visual sensors, thermal sensors, LiDAR, etc.) mounted on unmanned aerial systems to successfully detect both military and civilian objects in real time.

Coauthors: Domen Mongus, Mitja Žalik, Andraž Krivic, Andrej Mesner.

We present a vision-only approach for real-time camera pose estimation using eye-tracking glasses equipped with passthrough camera. The system extracts deep convolutional features with SuperPoint and matches them to a prebuilt 3D map of the vehicle interior, followed by PnP pose solving with RANSAC for robust localization. Implemented in a multi-threaded capture pipeline, it supports real-time processing for multiple passengers and visualizes ego-pose trajectories, demonstrated through a driver attention-tracking application in a JLTV simulator.

Autonomous robots operating in underwater and aerial environments without access to GNSS signals encounter significant sensing and localization challenges due to the inherent complexity of these environments. Traditional localization methods often fall short under such demanding conditions, emphasizing the need for robust yet cost-efficient solutions tailored specifically to these scenarios. In this talk I will overview the key challenges of deploying vision-based and acoustic localization approaches in realistic underwater and aerial settings. I will also highlight recent research trends and open questions in the field and give insights from our research at the Visual Cognitive Systems (ViCoS) laboratory focused on practical robotic localization in these challenging environments.

We will demonstrate the use of the ModalAI Starling 2 UAV for autonomous localization and real-time mapping of its surroundings using the SLAM (Simultaneous Localization and Mapping) method. The key advantage of this solution is its ability to accurately determine its position without relying on external navigation systems such as GPS, which is particularly crucial in military operations conducted in unpredictable or disrupted environments.Furthermore, the prototype incorporates artificial intelligence techniques—specifically, the YOLO neural network—to enable fast and reliable recognition of strategically important or potentially dangerous objects. As a result, the UAV automatically detects military vehicles, critical infrastructure, suspicious items, and other threats, transmitting the gathered information to command systems in real time.This approach facilitates early detection and continuous monitoring of changes on the ground, thereby enhancing security, increasing situational awareness, and providing an additional tactical advantage in the surveillance and response to aerial threats.

Marko Hölbl is an associate professor of computer science and vice dean of research at the Faculty of Electrical Engineering and Computer Science, University of Maribor (UM FERI). His research focuses on cybersecurity, cryptography, and privacy protection. He lectures at UM FERI, the Faculty of Criminal Justice, and the Faculty of Natural Sciences and Mathematics (FNM), covering information and network security and digital forensics topics. He has been a visiting researcher at TU Graz and INSA Orléans. Marko serves as Secretary General of CEPIS LSI and EAEEIE and a member of ECSO, SRIP GoDigital, and the Cybersecurity Section of the Chamber of Commerce and Industry of Slovenia. He is also a member of the Government Office for the Protection of Classified Information’s Advisory Working Group on Cryptography. Marko has actively participated in several research projects under Horizon Europe, Digital Europe, the European Space Agency, and other programmes.

Andrea Pompili, a cybersecurity specialist with 20 years of experience, began his career as a programmer. He transitioned to enterprise software development, then computer security, working on projects for Wind Telecommunication and Telecom Italia. Andrea led multiple cybersecurity initiatives for the Military and Government markets, diving into malware research, hybrid network attack and defence patterns, and cyber detection and response innovations. He has been part of the OWASP ZAP project. He is now the Chief Scientist Officer at CY4Gate, overseeing cyber intelligence and integrated Cyber/EW activities, contributing to multiple international cybersecurity working groups.

Milan Gabor is a recognised cybersecurity expert, ethical hacker, and founder of Viris, a company specialising in information security, penetration testing, and security awareness. With extensive experience in red teaming, threat analysis, and security consulting, he is a frequent speaker at international conferences and an active trainer in the field of cybersecurity. Milan is known for his practical, hands-on approach and dedication to improving organisational resilience against modern cyber threats.

Matej Rabzelj is a cybersecurity researcher at the Laboratory for Telecommunications, Faculty of Electrical Engineering, University of Ljubljana. His work focuses on cyberattack modelling, penetration testing, and deception technologies. He has been involved in national, EC, and EDA projects assessing cyber threats in critical infrastructures. He has co authored research articles and conference papers, developed security solutions, and participated in NATO cyber defence exercises.

Marko Hölbl is an associate professor of computer science and vice dean of research at the Faculty of Electrical Engineering and Computer Science, University of Maribor (UM FERI). His research focuses on cybersecurity, cryptography, and privacy protection. He lectures at UM FERI, the Faculty of Criminal Justice, and the Faculty of Natural Sciences and Mathematics (FNM), covering information and network security and digital forensics topics. He has been a visiting researcher at TU Graz and INSA Orléans. Marko serves as Secretary General of CEPIS LSI and EAEEIE and a member of ECSO, SRIP GoDigital, and the Cybersecurity Section of the Chamber of Commerce and Industry of Slovenia. He is also a member of the Government Office for the Protection of Classified Information’s Advisory Working Group on Cryptography. Marko has actively participated in several research projects under Horizon Europe, Digital Europe, the European Space Agency, and other programmes.