A team of German researchers is teaching drones to hunt for human screams. And while that sounds quite grim, nay terrifying, the technology is being developed for an extremely good cause.
The technology was presented by Macarena Varela at the annual meeting of the Acoustical Society of America this week. Varela is an engineer with the Fraunhofer Institute for Communication, Information Processing and Ergonomics (FKIE). She says that scream-seeker drones could prove extremely beneficial in post-disaster scenarios.
For example, if there is a catastrophe such as an earthquake and people are trapped under the rubble, every minute counts. However, given the scale of such disasters, even when rescue crews work around the clock to find survivors, the success rate is often low.
On the other hand, if drones that can detect screams are flying over a disaster-hit zone, they cover much larger areas in a short span of time. As such, they can enable better decision-making on the ground and guide first responders where to dedicate their efforts.
Decoding scream-seeker drones
The system works by equipping drones with a cluster of small microphones. But Varela is quick to point out that the project is still ongoing and requires a lot of testing and optimization. She explains:
We have already successfully detected and angularly located impulsive sounds very precisely near distances with the presence of drone noise. We will be testing the system on a flying drone to measure impulsive sounds, such as screams, and process the data with different methods to also estimate the geographical positions of the sounds.
But doesn’t the fact that the drone itself is not quiet, hamper its ability to pick up muffled sounds? Varela agrees that the number and distribution of microphones placed on the drones affect their listening capabilities significantly. But teaching drones to navigate particularly complex soundscapes is exactly what the team is trying to do:
We have a lot of experience in filtering noise, such as wind noise, extremely loud helicopter noise, ground vehicles noise, and more. We use different types of filters to be able to reduce noise, and we use diverse detection procedures to extract the signals of interest, such as impulsive sounds or screams.
FTC: We use income earning auto affiliate links. More.