At approximately 1:25 AM on June 24, 2021, the 12 story Champlain Towers South condominium in Miami, Fl came crashing to the ground. It only took 12 seconds for the building to be reduced to a pile of rubble. There was an immediate rush to evacuate the residents of the building’s 136 units. Soon, rescue crews would need to begin the devastating task of searching for victims, with the hope of finding survivors. On July 26, authorities were able to find the final missing person from the collapse, totaling 98 deaths.

More than 80 rescue units converged in Miami to help in the rescue efforts. Rescuers were lucky to be able to pull many survivors out from the debris. Unfortunately, after a certain time, they knew they would only be recovering bodies. Shortly before this tragedy, a researcher from the Fraunhofer Institute for Communication, Information Processing, and Ergonomics (FKIE) in Germany published a paper on how drones could be useful in just such dire circumstances. Written by Macarena Varela, and titled “Saving Lives During Disasters by Using Drones”, the study looked into how drones could use an auditory process to identify people calling out for help.

As Macarena points out at the beginning of her paper, in disaster scenarios, like earthquakes, every second counts when trying to find victims. But at the same time, the rescue teams need a way to search a site without posing risk to themselves, the buried victims, or others. In a situation like the Miami condo collapse, looking for victims by moving debris randomly could have caused people to become further trapped or injured due to the unstable environment. Macarena’s concept is to equip a drone with an acoustic system that FKIE is developing.

This system can filter out environmental noise, like traffic or the drone’s rotors, to identify specific kinds of sounds. “Survivors,” Macarena states, “typically plead for help by producing impulsive sounds, such as screams.” The acoustic system has an array of microphones and receivers that can identify noises such as screams or repetitive percussive sounds like someone banging for help. Once the system identifies where a scream is coming from, it automatically tells the drone to go to that location.

The system uses 32 MEMS (microelectromechanical system) as the microphone array. The MEMS are small in inexpensive, so Macarena says that they can play around with finding the right number of microphones to cover a wide area without weighing down the drone. “We opted for a very particular array called Crow’s Nest, where all microphones are randomly positioned in a sphere. This type of array provides sound coverage in every direction and [is] equally good in all directions,” Macarena explained.

Ideally, if Macarena’s drone concept works it could lead rescuers to the exact point where a victim is buried under rubble, such as what happened in Miami. She is still working on improving the number and positioning of the MEMS in the array. “In previous tests in the lab,” Macarena said, “we were able to detect impulsive sounds, such as clapping, having rotor noises present. We are currently processing the data with the drone flying.” The system hasn’t been perfected yet, but Macarena is confident that her team has the expertise to see drones with acoustic locators being used in search and rescue soon.