Slashdot Mirror
Engineers Teach a Drone To Herd Birds Away From Airports Autonomously (techxplore.com)
An anonymous reader quotes a report from Tech Xplore: Engineers at Caltech have developed a new control algorithm that enables a single drone to herd an entire flock of birds away from the airspace of an airport. The algorithm is presented in a study in IEEE Transactions on Robotics. Herding relies on the ability to manage a flock as a single, contained entity -- keeping it together while shifting its direction of travel. Each bird in a flock reacts to changes in the behavior of the birds nearest to it. Effective herding requires an external threat -- in this case, the drone -- to position itself in such a way that it encourages birds along the edge of a flock to make course changes that then affect the birds nearest to them, who affect birds farther into the flock, and so on, until the entire flock changes course. The positioning has to be precise, however: if the external threat gets too zealous and rushes at the flock, the birds will panic and act individually, not collectively.
To teach the drone to herd autonomously, Soon-Jo Chung, an associate professor of aerospace, and his colleagues [...] studied and derived a mathematical model of flocking dynamics to describe how flocks build and maintain formations, how they respond to threats along the edge of the flock, and how they then communicate that threat through the flock. Their work improves on algorithms designed for herding sheep, which only needed to work in two dimensions, instead of three. Once they were able to generate a mathematical description of flocking behaviors, the researchers reverse engineered it to see exactly how approaching external threats would be responded to by flocks, and then used that information to create a new herding algorithm that produces ideal flight paths for incoming drones to move the flock away from a protected airspace without dispersing it. The team tested the algorithm on a flock of birds near a field in Korea and found that a single drone could keep a flock of dozens of birds out of a designated airspace. The effectiveness of the algorithm is only limited by the number and size of the incoming birds.
To teach the drone to herd autonomously, Soon-Jo Chung, an associate professor of aerospace, and his colleagues [...] studied and derived a mathematical model of flocking dynamics to describe how flocks build and maintain formations, how they respond to threats along the edge of the flock, and how they then communicate that threat through the flock. Their work improves on algorithms designed for herding sheep, which only needed to work in two dimensions, instead of three. Once they were able to generate a mathematical description of flocking behaviors, the researchers reverse engineered it to see exactly how approaching external threats would be responded to by flocks, and then used that information to create a new herding algorithm that produces ideal flight paths for incoming drones to move the flock away from a protected airspace without dispersing it. The team tested the algorithm on a flock of birds near a field in Korea and found that a single drone could keep a flock of dozens of birds out of a designated airspace. The effectiveness of the algorithm is only limited by the number and size of the incoming birds.
I mean, what could go wrong?
=^..^= all your rodent are belong to us
I'm more interested in how long it'll take for the birds to learn that the drone can't actually hurt them, and that they can either ignore it, or attack it and get it out of the sky?
Read the paper, it seems that the issue is that the flocking birds have an exclusion zone when they're flying - when the drone enters it, the bird moves away and enters other birds exclusion zones causing a ripple effect and changing the direction of the flock.
So, this isn't something that birds will learn to ignore as not a threat.
Mimetics Inc. Twitter
Drone not effective against Rodan or any other Kaiju!
+1 insightful!
I have seen this happen with other bird deterring techniques.
This one seems immunized to this although but I have never seen it in use nor do I know if it is efficient:
https://www.birdbgone.com/prod...
Everything I write is lies, read between the lines.
It's worth going through the paper, this is some very interesting (and understandable) research with an actual application at the end. This is an study of flocking bird behaviour with the mathematics of the birds in the flock and the resulting flock behaviour being modeled and influenced using the drone.
I don't know how practical it will be for the thousands of airports out there but for large municipal airports i would think it's very valuable.
Mimetics Inc. Twitter
Seriously, there are some non-airport sites in my city that have gigantic murders of crows. It's even worse than a scene out of Hitchcock's "The Birds"
The city spends a ton of money driving them out with fireworks, blasted sound and lights, but they're back every year.
I'm more interested in how long it'll take for the birds to learn that the drone can't actually hurt them
Step one: Condition birds to believe that the drone is harmless
Step two: Mount a shotgun on the drone.
Agreed. Agricultural bird scarers use a 12v battery, propane gas cylinder
They also cover a very small area, and don't work well once the birds get used to them.
I have seen them used in cherry orchards near harvest time, and nowhere else.
To teach the drone to herd autonomously, Soon-Jo Chung, an associate professor of aerospace, and his colleagues [...] studied and derived a mathematical model of flocking dynamics to describe how flocks build and maintain formations, how they respond to threats along the edge of the flock, and how they then communicate that threat through the flock. Their work improves on algorithms designed for herding sheep, which only needed to work in two dimensions, instead of three. Once they were able to generate a mathematical description of flocking behaviors, the researchers reverse engineered it to see exactly how approaching external threats would be responded to by flocks, and then used that information to create a new herding algorithm that produces ideal flight paths for incoming drones to move the flock away from a protected airspace without dispersing it.
Sweet jesus, you fly back and forth around the area you don't want the birds to be at. They'll be pissed at the harassment and leave. Start in the middle, work your way out. Try to time it or send a signal so that the drone isn't exactly where you don't want flying things to be.
A lot of airports currently have a dude riding around on a golf cart with shotgun shell blanks and "seal bombs" aka basically quarter sticks of dyanmite that are meant to scare away seals but work for birds too. That is probably the #1 most sought after job (by males at least) in the entire world and eliminating it is simply unacceptable due to its dream-crushing for everyone aware of it.
Please stop implying there's some type of learning or AI going - there isn't. A "herding algorithm" reverse engineered form "mathematical description of flocking behaviors" is not a learned behavior. It's a programmed algorithm... computer code. Leaning implies an automaton developed some insight independently... Just like a self driving car will never know where left your sock in the bedroom... let alone understand the context of what a sock is....
Flocks of birds, schools of fish and herds of cattle behave in different ways.
The scientists and engineering who designed those drones are using patterns of behavior specific to flocks of birds, such as exclusion zones, which motivate them to change their course to avoid proximity to others. Anyone who has seen sheep huddling together will understand that what works for birds will not necessarily work for sheep.
Just because throwing a handful of corn makes cats dodge does not mean that chicken will do the same.
No good deed goes unpunished...
The main problem they have with systems designed to keep birds away from airports is that the birds quickly learn to ignore the system. This drone system also needs to learn how to continuously adjust it's behavior to still scare the birds just enough.
I've abandoned my search for truth; now I'm just looking for some useful delusions.
I fly drones at a local airfield and can vouch for the fact that they're a very effective method of getting rid of birds from such places.
When I first started flying at the airfield there were healthy (albeit not for the aircraft that used the place) populations of gulls, pluvers and (at certain times of the year) ducks. It was impossible to deter those birds using ground-based techniques but chasing them well beyond the airfield boundaries with a small racing drone has resulted in a dramatic drop in numbers.
Since I started driving the birds away (be still all you animal-rights activists) there has not been a single bird-strike at the airfield.
Of course the media would rather report that, as a drone flier, I could be spying on people, carrying high-explosives and trying to bring down airliners -- but then again we all know that what you read in the media is (these days) far from the truth.
We've been herding away birds from fields for several millenia. Should've asked a farmer, not an engineer.
Custom electronics and digital signage for your business: www.evcircuits.com
Can I use it on telemarketers?
Table-ized A.I.
The summary says they started with a 2D algorithm made for herding sheep, so yes, it has been done and has been mentioned.
sez you
Hitting a bird with the drone prop would most certainly break the prop and crash the drone.
You people don't even understand the problem. Birds in the flight path of jets get pulled into the engines and potentially blow up the engines, on landing or take-off, being the worst possible time for such a thing. So they built this drone to herd birds out of the way, awesome. But now that places a drone potentially in the flight path of jets, too. A drone made of plastics, metal and LiIon batteries would pretty much explode like a grenade inside the intake of a jet engine.
Now, if the inventors can tie the drone flight algos into the airport automated landing systems, then the drone should be able to calculate the vectors of jets with good accuracy, know when to herd birds around and which direction, and also know when to get itself the fuck out of the way. Potentially a big win, or potentially a really bad idea if it's not handled correctly.
=^..^= all your rodent are belong to us