Fraunhofer Institute for High Frequency Physics and Radar Techniques FHRWithin the UAV‑Rescue project, an international team of researchers is developing an AI‑assisted UAV‑based reconnaissance system to support emergency responders in complex disaster scenarios. The objective of the project is to enable faster and safer localization of trapped or missing persons, as well as to assist responders in situational assessment within hazardous confined environments, such as collapsed buildings, tunnels, or hard‑to‑access debris zones.
Within the project, we combine advanced radar, LiDAR, and AI technologies on autonomous aerial platforms. Using UAVs, indoor environments are explored semi‑autonomously, three‑dimensional situational maps are generated, and potential hazards are identified—without requiring responders to enter high‑risk areas. The integration of multiple sensor modalities enables robust environmental perception and human detection even under challenging visibility conditions.
Fusion of radar and LiDAR data enables precise mapping of unknown indoor mapping environments
A central research focus is the fusion of radar and LiDAR data for accurate 3D mapping of previously unknown indoor environments. Machine learning techniques are used to combine sensor data, analyze motion patterns, and develop autonomous search strategies. The AI system is guided by the experience of human responders and supports the semi‑autonomous navigation of UAVs within complex structures.
In addition, UAV‑Rescue investigates radar‑based methods for detecting vital signs. High‑resolution FMCW radar systems enable the detection of even minute movements such as respiration or heartbeat. A key scientific challenge, however, lies in compensating for the UAV’s own motion—which is significant relative to these vital signals—in order to achieve reliable human detection during flight. In this context, our researchers have, for the first time, successfully extracted human vital parameters using a radar sensor carried by a UAV.
Beyond the technical developments, the project also addresses ethical issues related to AI‑supported decision‑making in rescue operations. Key topics of the accompanying research include trustworthiness, accountability, and safe human–machine interaction.
The developed technologies are validated through extensive laboratory and field tests in collaboration with the Federal Agency for Technical Relief (THW) and international partners. The goal is to demonstrate an operational integrated system for intelligent situational assessment and victim localization in disaster scenarios.
Project Partners and Funding
The project involves seven German and eleven Austrian partners. The German partners include Fraunhofer FHR, Fraunhofer EMI, Leibniz University Hannover, Ruhr University Bochum, indurad GmbH, and the Federal Agency for Technical Relief (THW). The project is supported by associated partners including the German Rescue Robotics Center (DRZ), IMTEK, and Rosenbauer.
Funding was provided by the German Federal Ministry of Education and Research (BMBF; from 2025: BMFTR) with a total volume of €2,137,000.
Project duration: April 1, 2021 – July 31, 2023
