A new drone, developed by a German team, manages to hear human voices despite the noise of its six propellers and can go to the sound source thanks to its enhanced acoustic software and microphone. The person who pulls the drone over by making the sound hits the two pieces of wood together, making a more distinctive sound and providing the drone with more data it can use. In this way, the robot detects where the sound comes from and flies to the top of that area. As seen in the tests and the video below, the drone can successfully locate the researchers. The arrival of drone technology is a very promising sign for search and rescue efforts, but it's just as unfortunate for people who may have to hide from deadly drones and AI-powered autonomous weapons in the future.
The
sound-hunting drone was published under the title Saving Human Lives in
Disasters with a Drone and was developed at Fraunhofer FKIE, a research
institute in Germany. The drone was manufactured by the institute's Sensor Data
and Information Fusion Department, and its authors presented the research at
the Acoustic Society of America's meeting on June 8, 2021.
Important Breakthrough in Disaster Fight!
Drones and
many other robots are ideal tools to use when responding to a disaster. Sending
people to a newly collapsed building can put that person at risk; however, a
drone equipped with the right sensors can locate a person in danger and guide
disaster officials. This is an opportunity to load the "search" part
of search and rescue efforts on robots. Macarena Varela and Wulf-Dieter Wirth,
the authors of the article, explain:
Survivors
often impulsively cry out for help. For this reason, an acoustic system
developed at Fraunhofer FKIE is mounted on top of the drone so that it can
locate potential victims.
The work
of identifying and isolating screams is important and challenging. So the team
created a microphone array, attached it to a drone, and then added another
microphone that could pick up frequencies larger than other smaller microphones.
Programming the computer to detect screams meant filtering out other sounds,
such as the drone's roaring rotors, and converting the captured sound into
location data by combining values from the microphone array and determining the
direction of the screams.
According
to the news of the Washington Post, in the open area, the drone can detect a
person's location seconds after hearing his voice. Outdoor
tests are functional for sound calibration of such technologies; however, in
order to get full efficiency from the drone and to perform a good detection and
rescue mission, it needs to work in harsh environments with more noise.
If
researchers decide to develop the drones further, these sound detection tests
could be done in a controlled underground environment. In 2016, researchers in
France developed deep learning and artificial neural networks to detect screams
and loud conversations in the subway. The sound classifier created by the
project; He was able to distinguish screams and shouts and everything else.
Additionally,
as part of "DARPA's Underground Challenge," where teams sent robots
to courses to simulate search and rescue efforts, speakers embedded in thermal
dummies played human voice recordings at low volume to test the drones'
classification of sounds. The "Welcome to the Underground Challenge"
message from a lifeless body was pretty chilling.
In the
DARPA competition, sound is just one way for robots to detect human activity
underground. Drones can detect people in many different ways. For example, heat
emitted from thermal dummies can shine on infrared cameras, carbon dioxide can
be detected with chemical sniffers, signals from mobile phones can be picked up
with antennas, and they can also be detected with visual light optics cameras. In
rescue operations organized by the search and rescue team, searching for people
waiting to be rescued means discovering a variety of tools to detect human
lives, and then relying on rescuers.
The Same Drones Can Hunt Humans Too!
Because
DARPA-funded research comes from the Department of Defense, there's a veiled
detail about the robots it's developing to find humans: The same techniques
used by rescue teams to find people trapped underground can easily be applied
to soldiers performing search and clearing missions against armed enemies led
by robots.
It should
be clear that Fraunhofer FKIE's sound-hunting robot is not a military project,
and there is no data to suggest that the ultimate use of this robot will be for
military purposes. Robots that pinpoint the source of the sound are very useful
in rescue operations, just like in military operations. Even sound-capturing
robots can comfort an earthquake survivor waiting to be rescued.
Despite
their good side, reversible technologies can be dangerous. For example, a tool
designed for a harmless purpose can be easily programmed in other ways and used
for malicious purposes. Rescuers have valid reasons to be interested in the
sound-hunting drone. But they could easily develop robots like these that are
good at capturing and detecting sound for military and special purposes, like
killing or listening to people.
Comments
Post a Comment