"Robo Revolution," an excerpt from Frankenstein's Cat: Cuddling Up to Biotech's Brave New Beasts, by Emily Anthes
In the 1960s, the Central Intelligence Agency recruited an unusual field agent: a cat. In an hour-long procedure, a veterinary surgeon transformed the furry feline into an elite spy, implanting a microphone in her ear canal and a small radio transmitter at the base of her skull, and weaving a thin wire antenna into her long gray-and-white fur. This was Operation Acoustic Kitty, a top-secret plan to turn a cat into a living, walking surveillance machine. The leaders of the project hoped that by training the feline to go sit near foreign officials, they could eavesdrop on private conversations.
The problem was that cats are not especially trainable—they don’t have the same deep-seated desire to please a human master that dogs do—and the agency’s robo-cat didn’t seem terribly interested in national security. For its first official test, CIA staffers drove
Acoustic Kitty to the park and tasked it with capturing the conversation of two men sitting on a bench. Instead, the cat wandered into the street, where it was promptly squashed by a taxi. The program was abandoned; as a heavily redacted CIA memo from the time delicately phrased it, “Our final examination of trained cats . . . convinced us that the program would not lend itself in a practical sense to our highly specialized needs.” (Those specialized needs, one assumes, include a decidedly unflattened feline.)
Operation Acoustic Kitty, misadventure though it was, was a visionary idea just fifty years before its time. Today, once again, the U.S. government is looking to animal- machine hybrids to safeguard the country and its citizens. In 2006, for example, DARPA zeroed
in on insects, asking the nation’s scientists to submit “innovative proposals to develop technology to create insect-cyborgs.”
It was not your everyday government request, but it was an utterly serious one. For years, the U.S. military has been hoping to develop “micro air vehicles”—ultrasmall flying robots capable of performing surveillance in dangerous territory. Building these machines is not easy. The dynamics of flight change at very small sizes, and the vehicles need to be lightweight enough to fly, yet strong enough to carry cameras and other equipment. Most formidably, they need a source of power, and batteries light enough for microfliers just don’t have enough juice to keep the craft s aloft for very long. Consider two of the tiny, completely synthetic drones that engineers have managed to create: The Nano Hummingbird, a flying robot modeled after the bird, with a 6.5- inch wingspan, maxes out at an eleven-minute flight, while the DelFly Micro, which measures less than four inches from wingtip to wingtip, can stay airborne for just three minutes.
DARPA officials knew there had to be something better out there. “Proof-of-existence of small-scale flying machines . . . is abundant in nature in the form of insects,” Amit Lal, a DARPA program manager and Cornell engineer, wrote in a pamphlet the agency issued to the prospective researchers. So far, nature’s creations far outshine our own. Insects are aerodynamic, engineered for flight, and naturally skilled at maneuvering around obstacles. And they can power themselves; a common fly can cruise the skies for hours at a time. So perhaps, DARPA officials realized, the military didn’t need to start from scratch; if they began with live insects, they’d already be halfway to their dream flying machines. All they’d have to do was figure out how to hack into insects’ bodies and control their movements. If scientists could manage to do that, the DARPA pamphlet said, “it might be possible to transform [insects] into predictable devices that can be used for . . . missions requiring unobtrusive entry into areas inaccessible or hostile to humans.”
DARPA’s call essentially launched a grand science fair, one designed to encourage innovation and tap into the competitive spirit of scientists around the country. The agency invited researchers to submit proposals outlining how they’d create steerable insect cyborgs and promised to fund the most promising projects. What the agency wanted was a remote- controlled bug that could be steered to within five meters of a target. Ultimately, the insects would also need to carry surveillance equipment, such as microphones, cameras, or gas sensors, and to transmit whatever data they collected back to military officials. The pamphlet outlined one specific application for the robo-bugs—outfitted with chemical sensors, they could be used to detect traces of explosives in remote buildings or caves—and it’s easy to imagine other possible tasks for such cyborgs. Insect drones kitted out with video cameras could reveal whether a building is occupied and whether those inside are civilians or enemy combatants, while those with microphones could record sensitive conversations, becoming bugs that literally bugged you.
As far-fetched and improbable as DARPA’s dream of steerable robo-bugs sounds, a host of recent scientific breakthroughs means it’s likely to be far more successful than Acoustic Kitty was. The same advances that enabled the development of modern wildlife- tracking devices—the simultaneous decrease in size and increase in power of microprocessors, receivers, and batteries—are making it possible to create true animal cyborgs. By implanting these micromachines into animals’ bodies and brains, we can seize control of their movements and behaviors. Genetics provides new options, too, with scientists engineering animals whose nervous systems are easy to manipulate. Together, these and other developments mean that we can make tiny flying cyborgs—and a whole lot more. Engineers, geneticists, and neuroscientists are controlling animal minds in different ways and for different reasons, and their tools and techniques are becoming cheaper and easier for even us nonexperts to use. Before long, we may all be able to hijack animal bodies. The only question is whether we’ll want to.
DARPA’s call for insect cyborgs piqued the interest of Michel Maharbiz, an electrical engineer at the University of California, Berkeley. He was excited by the challenge of creating flying machines that merged living bodies and brains with electronic bits and bytes.
“What I wanted at the end of the day was a remote-controlled airplane,” Maharbiz recalls. “What was the closest thing to a remote-controlled airplane that I could get with these beetles?”
Maharbiz was an expert at making small electronic devices but an amateur when it came to entomology. So he started reading up. He figured that most scientists taking on DARPA’s challenge would work with flies or moths, longtime laboratory superstars, but Maharbiz came to believe that beetles were a better bet. Compared with flies and moths, beetles are sturdy animals, encased in hard shells, and many species are large enough to carry significant cargo. The downside: Scientists didn’t know much about the specific nerve pathways and brain circuits involved in beetle flight.
That meant that the first challenge was to unravel the insects’ biology. Maharbiz and his team began working with several different beetle species and eventually settled on Mecynorrhina torquata, or the flower beetle. It is a scary- looking bug—more than two inches long, with fearsome claws and a rhinoceros-like horn on the forehead. Through trial and error, the scientists homed in on a promising region of the beetle brain nestled at the base of the optic lobes. Previous research had shown that neural activity in this area helped keep the insect’s wings oscillating, and Maharbiz’s team discovered that when they stimulated this part of the brain in just the right way, they could start and stop beetle flight. When they sent a series of rapid electrical signals to the region, the beetle started flapping its wings and readied itself for takeoff. Sending a single long pulse to the same area prompted the insect to immediately still its wings. The effect was so dramatic that a beetle in mid-flight would simply fall out of the air.
After he discovered these tricks, Maharbiz was ready to try building the full flying machine. The flower beetle’s transformation began with a quick trip to the freezer. In the icy air, the beetle’s body temperature dropped, immobilizing and anesthetizing the insect. Then Maharbiz and his students removed the bug from the icebox and readied their instruments. They poked a needle through the beetle’s exoskeleton, making small holes directly over the brain and the base of the optic lobes, and threaded a thin steel wire into each hole.
They made another set of holes over the basalar muscles, which modulate wing thrust and are located on either side of the beetle’s body. The researchers pushed a wire into the right basalar muscle. Stimulating it would cause the beetle’s right wing to start beating with more power, making the insect veer left. They put another wire into the left basalar muscle; they would use it to steer the beetle to the right. The loose ends of all these wires snaked out of their respective holes and plugged into a package of electronics mounted with beeswax on the beetle’s back. This “backpack” included all the equipment Maharbiz needed to wirelessly send signals to the beetle’s brain: a miniature radio receiver, a custom- built circuit board, and a battery.
Then it was time for a test flight. One of Maharbiz’s students called up their custom- designed “Beetle Commander” software on a laptop. He issued the signal. The antennae jutting out of the beetle’s backpack received the message and passed it along to the circuit board, which sent electricity surging down the wire and into the beetle’s optic lobe. The insect’s wings began to flap. The empty white room the researchers used as an airfield filled wiTha buzzing sound, and the bug took flight. The beetle flew on its own— it didn’t need any further direction from human operators to stay airborne—but as it cruised across the room, the researchers overlaid their own commands. They pinged the basalar muscles, prompting the beetle to weave back and forth through the room, as if flying through an invisible maze. It wouldn’t have looked out of place going up against a stunt pilot at an air show. Another jolt of electricity to the optic lobe, and the beetle dropped out of the air and skittered across the tile floor.
As soon as Maharbiz presented his work, the news stories came fast and furious, with pronouncements such as “The creation of a cyborg insect army has just taken a step closer to reality,” “Spies may soon be bugging conversations using actual insects, thanks to research funded by the US military,” and more. A columnist speculated about the possibility of a swarm of locust drones being used as vehicles for launching deadly germs. There was chatter about beetles that had been “zombified,” and references to “the impending robots vs. humans war.”
When Maharbiz reflects upon this media frenzy, he admits that the immense public interest in his work doesn’t surprise him. The research, after all, is practically primed to light up the futuristic fantasy centers of our brains. Insects, even without modifications, seem like weird, alien organisms to many of us. As Maharbiz explains, “Insects have inherently some sort of strange, science fiction quality that a bunny doesn’t have.” Add in miniature electronics, flying devices, animal- machine hybrids, and covert military operations, and you have a recipe for dystopian daydreaming.
But Maharbiz bristles at the most sinister suggestions, at the media coverage that suggests his beetles are the product of, as he puts it, “some evil government conspiracy.” As for the possibility that the U.S. government is planning to use the bugs to build a killer insect army or to spy on its own citizens? “I think that’s nonsense,” he says. His beetles haven’t been sent out into the field yet—they still need some refinement before they’re ready for deployment—but if and when they are, Maharbiz says he expects his bugs to be used abroad, in routine military operations. (Of course, some people may find that “equally reprehensible,” he acknowledges.) There are civilian applications, too. Imagine, Maharbiz tells me, an army of beetle- bots, steered to the scene of an earthquake. The bugs could be outfitted with temperature sensors, guided through rubble, and programmed to send messages back to search teams if they detect any objects that are close to human body temperature; rescuers would then know exactly where to search for survivors.
Whatever the application, future insect commanders will have options that go beyond beetles. Maharbiz is working on a remote-controlled fly, which he anticipates being especially difficult to build. “The fly is so small and the muscles are so packed and everything’s so tiny,” he says, that even just implanting the electronics will be challenging. A Chinese research team has managed to start and stop flight in honeybees, and Amit Lal, the engineer who led the DARPA program, has created steerable cyborg moths.
One of Lal’s innovations has been figuring out how to take advantage of morphogenesis, the process by which many species of ate on a pupa than an adult insect. The procedure is so simple that it could enable the “mass production of these hybrid insect- machine systems,” the scientists wrote.
Still, the robo-bugs aren’t quite ready for their tour of duty. Our directional control is still pretty crude. Ultimately, we’ll want to do more than make an insect simply veer left . We’ll want to be able to command it to turn, say, precisely 35 degrees to the left or navigate a complicated three-dimensional space, such as a chimney or pipe. There’s also the matter of the surveillance equipment. So far, the main focus has been on building insects that we can steer, but for these cyborgs to be useful, we’ll need to outfit them with various sensors and make sure that they can successfully collect and transmit environmental information. And though the cyborg insects power their own flight—something that completely robotic fliers cannot do—the surveillance equipment will need to get its electricity from somewhere.
One intriguing possibility is to use the insect’s own wings as a source of power. In 2011, a team of researchers from the University of Michigan announced that they had accomplished just that by building miniature generators out of ceramic and brass. Each tiny generator was a flattened spiral—imagine the head of a thumbtack, if it were shaped from a tight coil of metal rather than a single flat sheet—measuring 0.2 inches across. When they were mounted on the beetle’s thorax, these generators transformed the insect’s wing vibrations into electrical energy. With some refinement, the researchers note, these energy- harvesting devices could be used to power the equipment toted around by cyborg bugs.
Insects could give us a cyborg-animal air force, zooming around the skies and searching for signs of danger. But for terrestrial missions, for our cyborg- animal army, we’d have to look elsewhere. We’d have to look to a lab at the State University of New York (SUNY) Downstate, where researchers have built a remote- controlled rat.
We’ve been rooting around in rat brains for ages; neuroscientists oft en send electrical signals directly into rodents’ skulls to elicit certain reactions and behaviors. Usually, however, this work requires hooking a rodent up to a system of cables, severely restricting its movement. When the SUNY team, led by the neuroscientist John Chapin, began their work more than a de cade ago, they wanted to create something different—a method for delivering these electrical pulses wirelessly. They hoped that such a system would free researchers (and rats) from a cumbersome experimental setup, and enable all sorts of new scientific feats. A wireless system would allow scientists to manipulate a rat’s movements and behaviors while it was roaming freely and give us a robo-rodent suitable for all sorts of special operations. Rats have an excellent sense of smell, so cyborg rats could be trained to detect the scent of explosives, for instance, and then steered to a field suspected to contain land mines. (The task would pose no danger to the animals, which are too light to set off mines.) Or they could be directed into collapsed buildings and tasked with sniffing out humans trapped beneath the rubble, performing a job similar to the one Maharbiz imagines for his cyborg insects. “They could fit through crawl spaces that a bloodhound never could,” says Linda Hermer-Vazquez, a neuroscientist who was part of the SUNY team at the time.
But before any of that could happen, the SUNY scientists had to figure out how to build this kind of robo-rat. They began by opening up a rat’s skull and implanting steel wires in its brain. The wires ran from the brain out through a large hole in the skull, and into a backpack harnessed to the rodent. (“Backpack” seems to be a favorite euphemism among the cyborg- animal crowd.) This rat pack, as it were, contained a suite of electronics, including a microprocessor and a receiver capable of picking up distant signals. Chapin or one of his colleagues could sit five hundred yards away from the rat and use a laptop to transmit a message to the receiver, which relayed the signal to the microprocessor, which sent an electric charge down the wires and into the rat’s brain.
To direct the animal’s movements, the scientists implanted electrodes in the somatosensory cortex, the brain region that processes touch sensations. Zapping one area of the cortex made the rat feel as though the left side of its face was being touched. Stimulating a different part of the cortex produced the same phantom feeling on the right side of the rat’s face. The goal was to teach the rodent to turn in the opposite direction of the sensation. (Though that seems counterintuitive, it actually works with the rat’s natural instincts. To a rodent, a sensation on the right side of the face indicates the presence of an obstacle and prompts the animal to scurry away from it.)
During the training process, the SUNY scientists used an unconventional system of reinforcement. When the rat turned in the correct direction, the researchers used a third wire to send an electrical pulse into what’s known as the medial forebrain bundle (MFB), a region of the brain involved in processing plea sure. Studies in humans and other animals have shown that direct activation of the MFB just plain feels good. (When the scientists gave the rats the chance to stimulate their own MFBs by pressing down on a lever, the animals did so furiously—hitting the lever as many as two hundred times in twenty minutes.) So sending a jolt of electricity zinging down to a rat’s MFB acted as a virtual reward for good behavior. Over the course of ten sessions, the robo-rats learned to respond to the cues and rewards being piped into their brains. Scientists managed to direct the rodents through a challenging obstacle course, coaxing them to climb a ladder, traverse a narrow plank, scramble down a flight of stairs, squirm through a hoop, and then navigate their way down a steep ramp.
As a final demonstration, the researchers simulated the kind of search- and- rescue task a robo-rat might be asked to perform in the real world. They rubbed tissues against their forearms and taught the rodents to identify this human odor. They constructed a small Plexiglas arena, filled it with a thick layer of sawdust, and buried human- scented tissues inside. When they released the robo-rats into the arena, the animals tracked down the tissues in less than a minute. The scientists also discovered that the rats that received MFB rewards found the target odors faster and dug for them more energetically than rodents that had been trained with conventional food rewards. As Hermer-Vazquez recalls: “The robo-rats were incredibly motivated and very accurate.”
Want more? You can pick up a copy of Frankenstein's Cat via Amazon
Excerpted from Frankenstein’s Cat: Cuddling Up to Biotech’s Brave New Beasts by Emily Anthes, published March 2013 by Scientific American / Farrar, Straus and Giroux, LLC. Copyright © 2013 by Emily Anthes. All rights reserved.