A Scalable Swarm Intelligence Algorithm for Autonomous UAV Search and Rescue Operations

This paper presents the design and implementation of an autonomous UAV-based search and rescue system developed within the Horizon Europe project P2CODE. The proposed system leverages a modular and scalable architecture integrating edge-based real-time video processing, AI-based human detection, asynchronous message communication, and persistent state logging, all orchestrated through a web-based operator interface. Central to the system is a swarm intelligence algorithm that partitions the search area among multiple UAVs, taking into account factors such as battery levels and initial positions to generate balanced and coherent flight paths. By combining a Divide Areas based on Robots' initial Positions (DARP) method with a Spanning Tree Coverage (STC) algorithm, the system ensures efficient and complete coverage of large outdoor regions. The operational workflow supports both fully autonomous exploration and reactive human-in-the-loop intervention in response to real-time detections. This work contributes a practical blueprint for large-scale, multi-agent coordination in dynamic and unstructured environments, advancing the state of the art in autonomous search and rescue missions.