New AI-enabled drone helps locate disaster victims by their screams
The world is only beginning to grasp the significance and applicability of unmanned aerial vehicles (UAVs) or drones in real-world situations. Adding to the growing list of such applications, researchers at Germany's Fraunhofer FKIE institute created a drone that can help locate trapped people by detecting their screams. The technology is intended to help first responders find victims after a natural calamity or disaster.
Using drones for post-disaster operations helps cover larger areas faster
The progress made on the prototype drone was showcased by one of the lead engineers, Macarena Varela, at an annual conference organized by the Acoustic Society of America. The Washington Post reported that using drones for post-disaster search-and-rescue operations enables responders to cover larger areas more easily and faster than manned efforts assisted by trained rescue dogs.
Drone can fly to areas where rescuers can't reach
Varela explained that if there's a collapsed building, the drone can alert and assist rescuers by flying to places rescuers can't get to by themselves and pinpointing trapped humans amid the debris. At present, drones are used to assess post-disaster damage to structures, use thermal imaging to detect heat signatures of trapped victims, and deliver critical supplies to those trapped in otherwise inaccessible regions.
AI matches microphone-detected sounds to pre-recorded samples to locate victims
The new drone relies on artificial intelligence (AI) to filter out ambient sounds and focus on detecting sound "signatures" similar to pre-recorded samples of screaming, tapping, and other sounds someone in need of help may make. The team reportedly uses microphones found on smartphones to accomplish the feat. We believe the system's hardware is fairly straightforward while the AI could get better over time.
Team has successfully conducted experiments, is now patenting location technology
So far, the team has successfully conducted "open field experiments" where the drone was able to identify a researcher calling out for help "within a few seconds." The engineers are now patenting their method of locating the source of sounds using microphones.
Drones, microphones still have limited operability in some scenarios
Varela said that the drone would be ideal for rescue operations after hurricanes, earthquakes, and wildfires. However, we believe that due to the inherent operational constraints associated with drones and the system's microphones, they aren't ideal for very windy or noisy conditions that may prevail in the wake of some disasters. Meanwhile, the team has reportedly been approached by several organizations with partnership proposals.
