An ant faced with two paths to a food source may not know which is best, but an ant colony can quickly sort this out: Individuals leave trails of chemicals called pheromones to mark the route from nest to food and back. As other ants follow, pheromones build up more rapidly along the shortest route, reinforcing the scent and guiding the colony en masse. 

A honeybee colony similarly pools its resources to select a home. Several hundred “scouts” head out to search, reporting back on potential sites. Other scouts follow up, choosing sides until a quorum has settled on a particular spot. This life-or-death decision draws on the collective wisdom — or swarm intelligence — of the hive. 

Inspired by examples like this, engineers are producing ensembles of small, insect-like robots that cooperatively perform jobs that might be difficult, dangerous or tedious for humans to carry out. As in nature, these robotic systems tend to be decentralized, composed of agents doing simple things that add up to communal achievements. Although the current price tags per unit are generally in the triple digits, developers hope one day to make the devices so inexpensive that some robots could be lost or damaged without jeopardizing the entire mission.

“In the end we always go back to nature to see what insects do,” says University of Washington engineer and robo-centipede designer Karl Böhringer. “They can do amazing things, and we want to be able to do them too.”

Doing It Like Ants

As an MIT college student in the early 1990s, James McLurkin decided to build robots that behaved like ants, an ambitious idea he attributes to “unbridled undergraduate hubris.” His critters were about an inch long. They had small internal computers, photoreceptors to help them avoid obstacles and two motors, one for locomotion and another for operating their mandibles. McLurkin got his charges to play follow-the-leader and tag. 

By 2004, as an MIT grad student, McLurkin mobilized 104 robots to find a hidden object in an abandoned Army building. The swarm located the target in repeated trials, with each individual maintaining contact with five or six nearby neighbors using infrared signals. 

Those successful tests got McLurkin thinking about real-world applications, such as searching for survivors after an earthquake or exploring the Martian surface. And today, as director of the Multi-Robot Systems Lab at Rice University, he’s aiming to do just that. Although his first robotic ants had antennae and mandibles, the r-one, McLurkin’s latest model, looks nothing like its predecessors or natural inspiration. Instead, it resembles an extra-large hockey puck. Each unit, powered by an internal lithium-polymer battery, costs about $250 — around 10 times cheaper than its more antlike ancestors. McLurkin hopes to get that cost down to $100 apiece, making large-scale deployments for search and rescue or other difficult missions more feasible.

RoboBees to the Rescue

In 2007, Robert Wood, the founder of Harvard’s Microrobotics Lab, became the first person to get a fly-size, flapping-wing robot to take off; a minuscule motor powered its plastic wings. The whole thing
— including a carbon-fiber body not much bigger than a fingernail — weighed just two-thousandths of an ounce. A colleague urged Wood to build a fleet of these vehicles, giving rise to the RoboBees project: a five-year, multidisciplinary venture involving biologists, computer scientists and engineers. 

The current RoboBee is considerably bigger, weighing nearly two-hundredths of an ounce. Its image-processing system tries to emulate the way real bees process images, with low-resolution camera “eyes” taking continual snapshots and forwarding information to the RoboBee’s “brain.”

One virtue is the minuscule weight of a single RoboBee. A firefighter, Wood suggests, could carry 1,000 of them in a pocket, weighing just a pound in all, and let them fly in or around burning buildings, dispatching information to rescue workers. Someday, RoboBees might track fast-changing events such as forest fires, oil spills or chemical plumes. Or they could pollinate flowers or crops — a pressing concern now that real bees are dying off from disease, chemical toxins and other causes. 

The robot’s biggest drawback, at the moment, is that it has to be wired to an external power source. Although achieving untethered, autonomous flight is a top priority, Wood doesn’t want to “wait until the perfect power source is available” before perfecting other elements of design. Some team members are trying to develop a lightweight power option, while others are working on sensors and manufacturing methods. “All these things are going on at once,” he notes. 

The engineers are also designing algorithms to regulate the behavior of thousands of RoboBees engaged in tasks like pollination. “We take our inspiration from nature,” says Harvard postdoc Karthik Dantu, “but then we decide what’s practical from an engineering standpoint.”

In 2007, Robert Wood, the founder of Harvard’s Microrobotics Lab, became the first person to get a fly-size, flapping-wing robot to take off; a minuscule motor powered its plastic wings. The whole thing
— including a carbon-fiber body not much bigger than a fingernail — weighed just two-thousandths of an ounce. A colleague urged Wood to build a fleet of these vehicles, giving rise to the RoboBees project: a five-year, multidisciplinary venture involving biologists, computer scientists and engineers. 

The current RoboBee is considerably bigger, weighing nearly two-hundredths of an ounce. Its image-processing system tries to emulate the way real bees process images, with low-resolution camera “eyes” taking continual snapshots and forwarding information to the RoboBee’s “brain.”

One virtue is the minuscule weight of a single RoboBee. A firefighter, Wood suggests, could carry 1,000 of them in a pocket, weighing just a pound in all, and let them fly in or around burning buildings, dispatching information to rescue workers. Someday, RoboBees might track fast-changing events such as forest fires, oil spills or chemical plumes. Or they could pollinate flowers or crops — a pressing concern now that real bees are dying off from disease, chemical toxins and other causes. 

The robot’s biggest drawback, at the moment, is that it has to be wired to an external power source. Although achieving untethered, autonomous flight is a top priority, Wood doesn’t want to “wait until the perfect power source is available” before perfecting other elements of design. Some team members are trying to develop a lightweight power option, while others are working on sensors and manufacturing methods. “All these things are going on at once,” he notes. 

The engineers are also designing algorithms to regulate the behavior of thousands of RoboBees engaged in tasks like pollination. “We take our inspiration from nature,” says Harvard postdoc Karthik Dantu, “but then we decide what’s practical from an engineering standpoint.”

Termite Builders

Although termites are best known for damaging wooden structures, they are among nature’s greatest builders. Less than an inch long, termites can create mounds as tall as 40 feet with an intricate network of tunnels inside. Equally remarkable is the fact that this massive construction effort proceeds without a “foreman.” Individuals collect mud and add it to a pile or repair holes or gaps, building a large and complex structure without any kind of central coordination.

Harvard computer scientist Radhika Nagpal and her colleagues in the TERMES Project are making artificial termites that can, like their natural counterparts, build structures much bigger than themselves. Each robot, about 7 inches long and about 4 inches wide, looks like an upside-down wheelbarrow supported by four “whegs,” hybrid wheel/legs. 

The robo-termite’s grippers can grab a brick, flip it onto its back and deposit it in the proper place. After successfully building a 10-brick staircase using a single termite, the team is gearing up to have three robots build a staircase of approximately 80 bricks, as well as more complex, castle-like structures. 

Rather than being told which brick to put where, a robot is programmed to reach a certain location. If that location is higher than its current position, the robot will keep laying down bricks, building a staircase until it can get there. No coordination with other robots is needed.

Harvard roboticist Nils Napp is developing modified versions of the termites that can pile up burlap rice sacks weighing a few pounds — a possible analog to reinforcing levees with sandbags or performing other construction jobs in places where it’s risky for humans, such as flood zones, contaminated environments, underwater or in outer space. 

Heavy Lifting With Centipedes

Although two feet work well enough for humans, engineer Karl Böhringer and his colleagues at the University of Washington wanted to see what a device could do with 512 feet. Their robotic centipede — built through microfabrication techniques similar to those used for making computer chips — is just over an inch long and about a third as wide, weighing less than two-hundredths of an ounce.

To achieve motion, the legs are built like sandwiches, with a layer capable of conducting electricity wedged between two layers of polymer. The electric current pulses 30 times a second, heating alternating layers of the sandwich with each surge. When heated, one polymer layer expands more than the other, causing the legs to alternately curve and straighten at a rapid clip. The robot can move forward, backward and sideways equally well.

One drawback is the unit’s speed — about 3 feet per hour. At that pace, it would take more than four days to travel the length of a football field. Böhringer plans to speed things up, perhaps by a factor of 100, though probably not fast enough to interest NASCAR fans.

The centipede’s greatest role might be as a miniature workhorse, since it can carry seven times its body weight. After trimming some excess fat, Böhringer believes the centipede could ultimately carry 50 times its body weight, which might make it useful for collecting samples on a Mars mission. That carrying capacity, if achieved, would put the robot on a par with ants — a rare instance where modern technology actually rivals nature.

RoboRoach Infestation

Cockroaches are even more despised than termites. Yet their ability to move quickly over virtually any terrain and get almost anywhere — crawling into small spaces, moving vertically or upside down — makes them not only formidable pests, but also an appealing model to roboticists. 

In 2009, University of California, Berkeley researchers unveiled the Dynamic Autonomous Sprawled Hexapod, or DASH. Their cockroach stand-in resembles a sled about 4 inches long and 2 inches wide propped up on six outstretched legs. The hexapedal design ensures stability; when DASH runs, its legs contact the ground in a tripod configuration.

Berkeley graduate student Paul Birkmeyer, who designed the original DASH, added claws to the cockroach’s feet. The modified robot, called CLASH, has interchangeable feet, including a magnetic option, enabling it to climb vertically on a range of smooth and rough surfaces, such as walls, couches or curtains. 

Another graduate student, Kevin Peterson, added motorized flapping wings and a tail to make DASH go faster, more than 4 feet per second. The modified unit also can scale steeper inclines and stay upright during falls. Peterson has dropped “DASH+Wings” from a variety of heights, and it almost always landed on its feet.

Looking to the future, the Berkeley team is devising applications for their hardy insects. “We could send them into places that are unsafe or where people simply can’t fit,” Peterson says. “The robots could look for explosives, check pipes for gas leaks or inspect bridges.” Cockroach pests go places we don’t want them to go. Now engineers are turning the tables, designing mechanized creatures that go places where people don’t want to go.