Flight Experiment of Indoor Drone Cooperative Mission Operation Assisted by Advanced AI-based Visual Navigation

Abstract—The evolution of autonomous systems for defense and security applications has increasingly emphasized cooperative multi-agent solutions capable of operating in complex and semi-structured environments. In this context, this work presents the experimental validation of the project named Swarm – APR – Artificial Intelligence (S.APR.AI), developed as part of a collaboration between the Air Force and academic researchers. The project addresses the critical challenge of autonomous target library generation and recognition in semistructured urban scenarios through the deployment of a heterogeneous multi-agent system composed of both unmanned aerial and terrestrial platforms. The validation campaign was conducted across three experimental phases: autonomous exploration, multi-modal cooperative target engagement, and dynamic aerial interception. Across these stages, the system demonstrated robust capabilities in target detection, classification, and tracking, leveraging electro-optical and infrared sensing. The results underline the effectiveness of cooperative swarm behaviors, real-time data fusion, and autonomou decision-making, even under stringent mission constraints. Overall, the outcomes confirm the operational readiness of cooperative aerial and terrestrial systems for surveillance, reconnaissance, and defense missions, while simultaneously providing a scalable and adaptable framework for future deployments of autonomous multi-agent systems.