How Fire Ants Avoid Traffic Jams
Worker ants keep the nest alive. They look for food, take care of the eggs, and dig all the tunnels. Fire ant colonies, for example, have hundreds of thousands of worker ants. You’d think traffic jams happen all the time. But they don’t!
The majority of the ants aren’t working, according to a study published in Science this week from researchers at the Georgia Institute of Technology. They remain idle to stay out of the way, leaving only 30% of the ants to dig a new hole. The researchers also believe the dynamic between idle and active ants could be applied to teaching small robots to dig together at an earthquake site or find shelter underground during a natural disaster.
Dr. Daniel Goldman, Dunn Family Professor of Physics at the Georgia Institute of Technology, joins Ira Flatow to discuss why idle ants keep traffic flowing.
Daniel Goldman is the Dunn Family Professor of Physics at the Georgia Institute of Technology in Atlanta, Georgia.
IRA FLATOW: This is Science Friday. I’m Ira Flatow. Worker ants keep the nest going. They look for food, take care of the eggs, and also dig an intricate tunnel system for the nest. In fire ant colonies, there can be thousands of ants moving around, trying to do their job.
And with all these ants coming in and out of the tunnels, now, traffic jams should be happening all the time, right? But they don’t. This is because more than half of the ants aren’t actually working. While it might sound like they are just lazy, they’re really just staying out of the way.
My next guest was curious about the perfect ratio of worker ants to idle ants and whether it could help swarm robots work in close quarters. Dr. Daniel Goldman is professor of physics at Georgia Institute of Technology. And his study appears in the journal Science this week. Welcome to Science Friday.
DANIEL GOLDMAN: Thank you for having me.
IRA FLATOW: So you’re a physicist, but you’re curious about how ants build tunnels?
DANIEL GOLDMAN: Correct. All my life I’ve been interested in animals. In particular, when I was a kid I liked lizards and snakes and even played with ants to some extent. But when I was training, I decided that it was more serious to become a physicist and study dynamical systems and pattern forming systems.
But in my later training, I learned that there were people who actually were interested in understanding the mechanics of organisms. And so for the last 12 years, I’ve combined those trainings to be a professor of physics at Georgia Tech, who’s group largely focuses on problems of organisms, including lizards, snakes, and, more recently, ants. with complex environments.
IRA FLATOW: So you had to actually learn how an ant builds a tunnel. How does it do that?
DANIEL GOLDMAN: Correct. Well, I should say right off the bat that we didn’t study tunnel/nest formation in natural environments. If you go out into the fields of Georgia or along the side of the roads, you’ll see these mounds, which basically house or cover the colony’s nest, which is a structure which can extend a meter deep into the ground, filled with ants and tunnels and brood and queens.
And we couldn’t study that in the laboratory in the way that a physicist likes to study things, which is to make careful, controlled situations. So basically, we would go and dig up nests in the ground and bring them back to the laboratory and flood them out.
These ants have a very interesting behavior, which features in kind of the behaviors I’ll discuss, in which if they are subject to floods, which happens where they evolved in the Pantanals of South America, the whole colony will pick up and raft down the river, forming a hydrophobic, basically, structure out of ant bodies. My colleague, David Hu, at Georgia Tech has studied this extensively. And when they hit dry land, then they have to get into the ground quickly because the nest is basically their shell of the super-organism.
IRA FLATOW: So how did you discover that some of the ants are sort of lazy? And that weren’t really lazy, there was a natural course they had to take?
DANIEL GOLDMAN: Yeah. Well, we decided to do careful, controlled experiments. We would take about 30 ants. And we chose 30 because we had to tag the ants to see who was doing what, where, and when.
And we had tried a number of techniques, including little paper barcodes glued to their abdomens, and the glues would just kill them. So a postdoc in my group, Dr. [? Dari ?] [? Monakhova, ?] figured out a clever technique to paint the ants with about 30 different color combinations using oil-based markers.
And she then put the ants into a little container filled with soil and made sure they dug a little tunnel near a clear sidewall. And then, basically, we tracked for hours and hours of video, which ants were coming to the tunnel, whether they had excavated soil, scraped it away from the tunnel face, and whether they were transporting that soil up to the surface.
IRA FLATOW: And so they figured out, in their own way, because that’s what they do, how to then delegate the work? Which ones are going to hang back so they don’t clog up the tunnel.
DANIEL GOLDMAN: Yes, and I should say right off the bat that we don’t know how they make those decisions. So the origin of the decisions by which the and say, well, the one we painted purple-blue is going to dig the most– we don’t know. Simply recording the activity or the number of visitations from different ants resulted in sort of a funny distribution, where you had basically 3 to 5 out of the 30 doing the bulk of the labor over 12 or 24 hours, and the rest doing little bits. Although about 50%, as you alluded to, we never saw visit at all.
IRA FLATOW: Wow, so why don’t they just make the tunnel wider so more ants [INAUDIBLE]?
DANIEL GOLDMAN: That is a terrific question, and we don’t know why. We have some evidence that the narrow tunnels allow them actually to climb up and down quite quickly while missing footsteps, because when they pitch back– which they do frequently– they encounter another tunnel wall with their antennae or heads, and they just get a little bump and go right along their way. So there seems to be some locomotion benefits to it.
IRA FLATOW: Right.
DANIEL GOLDMAN: We also think that making a narrow tunnel allows you to get deep into the ground quickly. So that’s basically what we know about that. And by the way, you can try to force the ants to make larger tunnels. You can start a hole with a larger diameter, and they always make it more narrow. Or if you start the hole with a narrow diameter, they always widen it. So they like to dig these tunnels of this diameter. And that is independent of whether it’s–
IRA FLATOW: Interesting.
DANIEL GOLDMAN: –Georgia clay or sandy stuff.
IRA FLATOW: Yeah. But you’re trying to apply this research to a bunch of swarm robots?
DANIEL GOLDMAN: Yeah.
IRA FLATOW: You’re going to teach the robots what you’ve learned about the ant?
DANIEL GOLDMAN: Well, in some sense, yes. I’m a physicist. And so the kind of studies we like to do are careful, biological studies, coupled to modeling studies. And so in our paper, we had computational models, called cellular automata models, which allow us to get some insight in silico as to what the ants might be thinking and why.
But it’s more satisfying to conduct what we call physical modeling studies, where we actually build robots– autonomous robots. And because we were physicists, we couldn’t build thousands of robots like the ants. We could build a few. But that turned out to be OK because the real story happens to be in these narrow tunnels– the kind of clogging and clustering that occurs when few numbers of ants or ant robots all get in each other’s way.
IRA FLATOW: So you can apply this to swarming robots and teach them not to get in each other’s way? Or if you don’t know how the ants decide what to do, how do you get the robots to know what to decide what to do?
DANIEL GOLDMAN: Well, it’s an excellent question. And what we did was, in fact, we programmed the ants to model what the ants were doing– programmed the robots the model what the ants were doing, in the sense that, when we first started this project, we were curious as to whether we could make groups of robots excavate a model system well.
And the student who was working on it, a guy named [? Vadim ?] [? Linovic, ?] had the idea, well, I’ll make a couple autonomous robots, and I’ll just send them in to that tunnel, and they’ll start to excavate. And indeed, one robot did pretty well. Two robots did better. Three even better.
But by the time he got to four, the whole excavation came crashing to a halt. And he tried all sorts of clever things, like rules which said, if a robot’s in a jam, it should wiggle a little bit or back up. And nothing was robust.
And then we realized what the ants were doing. And we tried these strategies in the robots. And very broadly speaking, combinations of the strategies, including staying out and giving up when the tunnel was too clogged, produced real benefits–
IRA FLATOW: Wow.
DANIEL GOLDMAN: –in the robot excavation performance in the laboratory.
IRA FLATOW: Do you ever bring in EO Wilson as a consultant on this?
DANIEL GOLDMAN: (CHUCKLING) No, we did not, although I would love to know what he thinks.
IRA FLATOW: [LAUGHS] (WHISPERING) You can give him a call. I think we have his number.
DANIEL GOLDMAN: [LAUGHS]
IRA FLATOW: (CHUCKLING) So where do you go from here with this? What have you learned, and how do you apply that to what you want to do?
DANIEL GOLDMAN: Well, you know, so I’m a scientist– a physicist. And so for me, it’s the joy of discovering secrets of nature. And these ants are just so incredible at what they do, that I feel like we’ve understood a little bit about how a group which has no leader and has limited information about what everyone else is doing can accomplish a complex task that no individual can.
We are presently trying to understand how the ants develop this optimal digging strategy. We have evidence in our computer models and robot models that, in fact, such a strategy can evolve pretty quickly.
One of the interesting experiments we did was take, in bucket of 30 ants, the top five excavators out and let the remaining 25 excavate a fresh batch. And we found that the remaining 25 dug just as effectively, but five more stepped up. So how that happens, we’re very interested to know– and how we can teach the robots to quickly converge on this distribution is interesting to know as well.
IRA FLATOW: These were fire ants you were using?
DANIEL GOLDMAN: Correct.
IRA FLATOW: Ugh. Were you a little careful about that?
DANIEL GOLDMAN: We’re very careful about that.
IRA FLATOW: [LAUGHING]
DANIEL GOLDMAN: We wear gloves. And the sides of the containers that we house them in are coated in baby powder because they can’t walk across it very well, climb across it. But if you don’t bother them, like most animals, they don’t tend to bother you. And they’re just great study subjects because they see soil and they dig.
IRA FLATOW: Is this because you’re in Georgia? Could you do this with other kinds of ants that might be found around the country?
DANIEL GOLDMAN: Yeah, we could do this– there are a lot of ant species that dig subterranean structures. And they’re quite beautiful. In fact, you can buy casts of these things– molten casts are made, particularly by Professor Walter Tschinkel, Florida State. And you can buy these casts and see the different structures. But fire ants turn out to be very convenient in Georgia because they’ve taken over everywhere.
IRA FLATOW: Do you think termites build the same way?
DANIEL GOLDMAN: That’s a great question. I don’t know.
IRA FLATOW: Yeah. Because they build those great tunnels and mounds and things.
DANIEL GOLDMAN: Correct. Correct, they build structures and tunnels. And I think that it’s well worth looking into.
IRA FLATOW: What about applying your bots to taking them to Mars or to another planet to work on their own digging stuff?
DANIEL GOLDMAN: Yeah. Well, I think this brings up an interesting point in that we presently have now robots that are pretty good flying robots– you know, the little drones that you can buy for a couple hundred bucks. And you can actually get swarms of those to do pretty interesting things, like if you remember the Olympics where there were swarms flying.
But we don’t have– and not even close to having– comparable robots in terrestrial environments, like rubble, like rock piles, like the material after a building collapses or an earthquake. And so we need to get robots that are actually capable of moving in those terrain.
And once you have robots that are capable in moving those terrain, you could imagine putting many of those robots into a nasty terrain. And once you have many, then you have a swarm. And then we think some of the principles we’ve discovered will be useful for such swarms.
IRA FLATOW: Well, we wish you great luck, Dr. Goldman.
DANIEL GOLDMAN: Thank you,
IRA FLATOW: Yeah. I saw that movie, The Swarm.
DANIEL GOLDMAN: [CHUCKLES]
IRA FLATOW: Read the book. [LAUGHS]
DANIEL GOLDMAN: (LAUGHING) Ditto.
IRA FLATOW: [LAUGHS] Daniel Goldman is professor of physics at the Georgia State Institute of Technology. Thanks again.