To optimize its delivery drones, maybe Amazon could take inspiration from the actual Amazon.
Army ants in Central and South America aggressively seek out the shortest path over the forest floor to bring home enough food and ensure the future of their colony. This focus on efficiency led the insects to develop a clever trick: They link their bodies together to fill potholes and build living bridges.
As more ants join in, the bridges shift locations to span larger and larger gaps, shortening the path ants have to take when carrying food back to the nest. But because each brick in the bridge is also a lost forager, the ants reach a point where a slightly better shortcut just isn’t worth the cost, according to new analysis of this insect construction work.
“Overall, that cost-benefit tradeoff is reached, but without any ants really knowing,” says study leader Chris Reid of the University of Sydney.
Reid’s study, appearing this week in PNAS, is the closest look yet at the architectural algorithms army ants use when they build bridges. Understanding these rules could help scientists design smarter robotic swarms, for instance, by programming self-assembling materials to create dynamic structures as big as life rafts or as small as surgical stents.
To see their subjects ant-scaping in the wild, Reid’s team headed into the jungle of Panama’s Barro Colorado Island. Army ants from the genus Eciton, though voracious little murderers, are prudent when it comes to sustainable hunting. After a hard day pillaging larvae from the colonies of other ants and wasps, they pick up and march to new territory a few hundred feet away.
“You come back the next day to where you had previously found these ants, and they’d be gone,” Reid says. The only way to reliably find them again was to catch the move in progress, which meant going into the jungle at night.
“So that was always a pretty fun experience—tarantulas everywhere, rumors of jaguars stalking the island, and all sorts of things like that” he says.
After marking the ants' new hunting grounds, the researchers would head back to camp and return the next day to find tight columns of raiders streaming along impromptu roadways between the temporary army ant nest and the nests of their victims.
The ants navigate using pheromones, so the team could take marked-up sticks from the path to use as road signs and re-direct traffic into their experiment. On the forest floor, they laid down white boards with a crook in the path shaped like an open triangle.
When Reid’s team recorded the action, they saw the ants problem-solving in real time. First a single ant stumbles her way over the one-body-length gap just under the crook and sticks in place. Then another ant, walking over her, lays down pheromones on the shortened path.
Soon, ants using the shortcut freeze in place to become part of the bridge, since frequent contact with other ants makes them more likely to lock in. As the bridge thickens, travelling ants prefer to walk farther from the crook, because that path is slightly shorter.
Increased traffic on the favored edge makes that side of the bridge grow as new workers join the architecture. At the same time, workers on the unpopular edge are rarely touched and start leaving. With time, the whole bridge starts to migrate away from the crook.
But every time they recorded the ants, Reid’s team saw the bridge stop shifting at some point in the middle of the gap.
“Why do they stop then?” he says. “You would imagine the process would continue all the way down, until they have this nice straight trail that goes over all the gaps in their environment.”
Zooming out to the colony level, the strategy makes good fiscal sense, the team thinks. A bridge can save time, but every worker caught up in one is also a worker not carrying food back to the nest. Once too many workers are off the road, further improving a bridge is a waste of precious resources.
“I would have just expected them to make the bridge that makes the shortest possible path,” says Georgia Tech’s David Hu, who has previously researched the living rafts that fire ants build during floods. “How do they know that this is the best bridge for them?”
Though it’s still unclear, Reid’s favored explanation is that the bridge stops shifting when the decrease in traffic becomes noticeable to the living structure. As the longer bridge sucks more ants off the road, the touches that prompt an ant to donate its body to the bridge become less common.
While Hu thinks this explanation is too rough to consider the mystery solved, he stresses that this “beautiful experiment” is a first step in understanding this kind of problem-solving behavior and eventually applying it to swarming robots.
“We have nothing built out of robotics that has this combination of moving really fast and also becoming building material,” he says. “They go between the walking state and the bridge-building state so quickly that this thing seems to just morph.”
In the future, Reid’s group plans to work with Harvard computer scientist Radhika Nagpal, who thinks the kind of thinking, calculating architecture that army ants are capable of would be useful for small, expendable robots in dangerous rescue operations. “They could self-assemble into larger structures—bridges, towers, pulling chains, rafts,” she says.
Beyond such technical applications, the ants themselves demonstrate the power of a leaderless but well-programmed swarm.
“They are a super-organism for sure,” Nagpal says. “I don’t see how one can go wrong being completely fascinated with how a group that big can do so much so quickly and without politics, and without hierarchies of managers and CEOs.“