Researchers Reference Flocking Birds to Improve Swarmbots

inghamb87 writes "Scientists have studied flocks of starlings and cracked the mystery behind the birds' ability to fly in large formations, and regroup quickly after attacks, without getting confused and ramming into each other. While the information is cool, some scientists seem to think that the best use of this knowledge is not to aid our appreciation of nature, but to make more effective robot swarms. We've talked about swarming robots many times before, but usually researchers look to insects for inspiration."

Comments

Boids

[Joaz Banbeck]

Craig Reynolds was doing this many years ago: http://www.red3d.com/cwr/boids/

Re:Boids

[TheBrakShow]

Perhaps most relevant to this article is this particular simulation made last year which actually demonstrates flocking birds.

http://www.dcs.shef.ac.uk/~paul/publications/boids/index.html

You can even play with the settings panel on the right side and set off "gunshots."

But yeah, this stuff is far from news.

Re:Boids

[mrogers]

The cool thing about this new model is that each bird only needs to track a fixed number of neighbours (seven in the starling flock on which the paper is based). IIRC every bird in the Boids model needs to track every other bird to keep the swarm cohesive.

I haven't read the paper yet, but it seems like there could be a parallel with gossip protocols and flooding protocols: if each bird tracks a small number of randomly chosen neighbours, information can move through the swarm just as efficiently as if each bird tracks every other bird.

Modelled after birds?

[Malevolent Tester]

Researching bird flight and it's applications: £2m
Developing autonomous swarming robots: £5m
Watching your prototype robots fly straight into the nearest window at high speed and die: Priceless

Odd to dismiss it so early

[StaticEngine]

Perhaps the OP could consider that not all robots are human killing machines, and this kind of swarming/flocking behavior could be applied to something like vehicular safety. I've often pondered the idea of lateral lines on fish, and how quickly a school of fish can become aware of the motions of surrounding fish and other obstacles, remaining in formation but moving as seemingly one unit. How great would it be if robotic cars could react thousands of times faster than a human, and in concert, to flow seamlessly around a tire blowout, or debris that fell off a truck onto the highway? Aren't these kinds of goals the very reason we do this kind of research, and isn't the application of this reserach to improve our quality of life the very thing that pushes mankind forward intellectually?

Re:Odd to dismiss it so early

[flyingsquid]

Perhaps the OP could consider that not all robots are human killing machines, and this kind of swarming/flocking behavior could be applied to something like vehicular safety. I've often pondered the idea of lateral lines on fish, and how quickly a school of fish can become aware of the motions of surrounding fish and other obstacles, remaining in formation but moving as seemingly one unit. How great would it be if robotic cars could react thousands of times faster than a human, and in concert, to flow seamlessly around a tire blowout, or debris that fell off a truck onto the highway?

Yeah, sure, like THAT will make a cool premise for a sci-fi movie...

FA Just Another Dreamy Blog

[foxylad]

I was intrigued about what the actual algorithm used by the starlings was, but the referenced article didn't elucidate. Eventually I found a link to http://www.telegraph.co.uk/earth/main.jhtml?xml=/earth/2008/01/29/scistarling129.xml hidden at the bottom - it has a little more detail. Enjoy!

jumbo packet swarms in mesh networks

[G4from128k]

I would imagine that some clever network folk could use this research to develop interesting parallel-distributed network management algorithms. After all, a large data packet is not unlike flying bird that does not "want" to collide with other packets in large network (= transport medium = "air"). Assuming the coordination packets are much much smaller than the data packets, this scheme would cost-effectively prevent collisions and congestion by optimizing the spread of data both cross-sectionally and longitudinally in a network.

Tag

[TurinPT]

Stop using whatcouldpossiblygowrong for crying out loud, it completely defeats the purpose of having tags if all the articles have the same tags.

Re:Tag

[gardyloo]

Perhaps something has gone wrong with the tagging system.

Re:Tag

[quest(answer)ion]

it completely defeats the purpose of having tags if all the articles have the same tags. making the same joke over and over will kill it regardless of the way the joke is made.

of course, /. is living proof that this stops no one.

Re:Tag

[Bongo Bill]

Stop using that tag? Yeah. That's a great idea. What could possibly go wrong?

Re:Link to article

[Bazzargh]

No, the actual article is here:

Interaction Ruling Animal Collective Behaviour Depends on Topological rather than Metric Distance: Evidence from a Field Study

Numerical models indicate that collective animal behaviour may emerge from simple local rules of interaction among the individuals. However, very little is known about the nature of such interaction, so that models and theories mostly rely on aprioristic assumptions. By reconstructing the three-dimensional position of individual birds in airborne flocks of few thousands members, we prove that the interaction does not depend on the metric distance, as most current models and theories assume, but rather on the topological distance. In fact, we discover that each bird interacts on average with a fixed number of neighbours (six-seven), rather than with all neighbours within a fixed metric distance. We argue that a topological interaction is indispensable to maintain flock's cohesion against the large density changes caused by external perturbations, typically predation. We support this hypothesis by numerical simulations, showing that a topological interaction grants significantly higher cohesion of the aggregation compared to a standard metric one.

Re:MOD PARENT UP

[Bazzargh]

Modelling bird swarming behaviour isn't new. ...getting it right is. If you rtfa, and Craig Reynolds work, you'd know that the boids simulation assumed that birds interacted with all nearest neighbours within a certain distance. The paper this article refers to proves by observing starling flocks that that isn't true - in fact the starlings interacted with the nearest 6 or 7 independent of the distance apart the birds were.

Re:its undeniable and sad

[MightyYar]

get over it, most of the good roboticists are or will be part of some weapon industry. Indeed, it's very hard to be part of an industry that doesn't at least indirectly help the military. Even an improvement in textile loom speed can produce cheaper uniforms. The military has a keen interest in everything from alternative fuels, to advances in materials science, to food preservatives. If you can think of an improvement for something, it is likely that it can (and will) be used in some way, however small, to kill people more efficiently.

As you say, get over it. :)