Open-Source Drone Intelligence for Faster Search and Rescue

Drones have become indispensable tools for search and rescue operations, offering rapid aerial coverage, low operational costs, and the ability to reach dangerous or inaccessible areas. However, current systems often rely on proprietary platforms that limit flexibility and scalability. Researchers from the University of the Bundeswehr Munich have developed AUSPEX , an open-source decision-making framework designed to coordinate fleets of unmanned aerial vehicles in complex emergency missions.

AUSPEX integrates artificial intelligence, autonomous planning, and human supervision in one modular software ecosystem. It enables drones from different manufacturers to work together seamlessly, plan flight paths, detect victims, and adapt their missions in real time. Built entirely on open standards, AUSPEX combines perception, reasoning, and execution capabilities using advanced AI methods and realistic virtual environments for training and validation.

Its transparency and interoperability make it a foundation for next-generation emergency-response robotics, where multiple drones collaborate intelligently to save lives in the first critical minutes after a disaster. AUSPEX—Automated Unmanned Aerial Swarm System for Planning and Execution—is a holistic AI-driven framework that enhances the decision-making capabilities of drone teams. It operates through five interconnected modules that mirror human cognitive processes: knowledge acquisition, perception, planning, execution, and onboard control.

The system’s knowledge base continuously updates environmental data from drone sensors, while a perception module interprets camera feeds using open-source image-recognition models such as YOLOv8 for detecting people or vehicles. Its planning engine combines classical AI planning, heuristic optimization, and large language model (LLM)-based reasoning to create efficient flight strategies even in unpredictable conditions. The execution module translates plans into real-time actions, coordinating drones through the ROS 2 communication backbone, a standard in robotic software.