Lieutenant Bug: Cyborg Insects Report for Duty (Berkeley Science Review 5/2010)
Those already skittish of insect creepy–crawlies will not be thrilled to discover that these six–legged creatures may soon
be outfitted with more outlandish features. Indeed, these tiny animals could become directed vehicles of human technologies, including surveillance cameras, sensors, and chemical or biological weapons. Researchers in the Department of Electrical Engineering and Computer Science are fashioning insects—beetles, to be precise—with wireless transceivers and neural implants so that a human operator can dictate where the beetles fly.
That’s right. These beetles can be flown with a hand–held control, much like hobby airplanes are flown with a joystick. But these are no children’s playthings. The agency funding this research, the Defense Advanced Research Projects Agency (DARPA), sees these insects as an important step toward the future, a future in which international espionage and warfare can be outsourced from error–prone humans to highly controllable robots and cyborg animals. The funding program, named Hybrid Insect Micro–Electro–Mechanical Systems (HI–MEMS), aims to take rein over insect locomotion, charging teams with the task of guiding an insect to a target with a radius of five meters from 100 meters away.
Professor Michel Maharbiz and his group are not far from satisfying these goals. They published their work last year in Frontiers in Integrative Neuroscience, generating a wild proliferation of media attention, from national news sources to personal blogs. Already, the beetle has been featured as one of TIME Magazine‘s 50 Best Inventions of 2009 and the MIT Technology Review‘s 10 Emerging Technologies of 2009. All of this press coverage has generated a hubbub of reader opinions ranging from moral outrage (“this is just plain wrong!”) to wry sarcasm (“I have already rendered this technology useless by developing countermeasures in the form of a rolled up newspaper”).
No doubt something like military–funded neurologically controlled beetles may incite some controversy. But those who think that piddling with the stuff of life is intrinsically wrong may want to reconsider historical context. “The bioengineering is not the frightening part. After all, we’ve been domesticating animals forever,” says Professor Jake Kosek of the Department of Geography, whose research focus lies at the intersection of nature, technology, and politics. “The real question is how these military–funded technologies will be used in the future,” Kosek says. And that may not be such a straightforward question to answer.
Free flight is one of nature’s marvels that engineers drool over. Yes, we can ferry millions of passengers around the world each day in
fat passenger planes and fly fighter jets in twirling backflips to dodge missiles. But try scaling an air vehicle to the size of an insect
and you will hit major roadblocks to achieving sustained stability and maneuverability. Many engineers, including a handful at UC Berkeley (see “Robot Flea Circus,” BSR Fall 2007) are trying to combine inspiration from nature with engineering finesse to create fully synthetic mechanical insects capable of flying and jumping. However, even the most successful examples of this have achieved mere seconds of sustained flight. Although these avenues of research have immense long–term implications in understanding the details of fine mechanical control needed to generate flapping–wing flight, their near–term applicability is limited.
Engineers like Maharbiz are instead turning their eyes towards using natural systems as starting platforms themselves. This road is also fraught with challenges, not the least of which is the choice of an appropriate insect host. According to Harvard biologist E.O. Wilson, the earth contains an estimated 10 quintillion (1019) insects, making up approximately eight million distinct species. The sheer diversity of these insect types complicates the selection of an appropriate species. To limit the choices, Maharbiz’s team made a clever decision based on flight mechanics.
Insect flight falls roughly under two categories: synchronous and asynchronous. Synchronous flight, as seen in dragonflies, grasshoppers, and moths, is characterized by a direct correspondence between neural stimulation and wing beats. Each beat of the wing is triggered by an individual neural impulse. Asynchronous flight, in contrast, uses neural impulses to trigger the wing muscles to oscillate for many beats. These wing muscles have a resonant muscle–mechanical system, allowing the wings to flap faster than the muscles contract, thus requiring fewer muscle contractions for more flaps. This is akin to a child’s swing set, in which a single push can keep the swing going for many cycles. Beetles and other insects with highwing beat frequencies possess asynchronous flight muscle systems.
While other HI–MEMS–funded teams were working with moths, Maharbiz reasoned that insects with asynchronous muscle systems would provide perfect platforms for low–power engineering. The motor neurons fire at much lower frequencies than the wing beat frequency, implying that the neuronal stimulus serves only to switch flight on and off, and to specify how quickly the beetle’s wings are beating. Thus, if a researcher were trying to manually control flight, it would be possible to trigger a beetle brain only occasionally, while a synchronous flier would require stimulation for every wing beat. The muscles could subsequently be triggered independently to achieve turning.
The easiest beetle for the team to acquire was Zophobas morio, the darkling beetle, because their larvae are often used as pet food. There was just one glitch. “The problem was that the beetle naturally doesn’t fly,” says Hirotaka Sato, the postdoctoral researcher leading the project.
“We did find one paper,” says Gabriel Lavella, another researcher in the Maharbiz lab, “showing that Zophobas could be stimulated to fly briefly,” which inspired the group to continue testing them. However, it soon became clear that these beetles were vastly underqualified for the job, even though Sato was able to scare them into flying for about five minutes. “I think I set the world record,” he quips.
Within several months, they were able to acquire a real flying beetle: the Cotinis texana beetle, which was readily supplied from horticulturalists in Florida looking to get rid of these garden pests. Although the neuroanatomy of insect flight control is still not well understood, the literature indicated that visual and auditory stimuli could sometimes be sufficient to initiate flight, which gave the researchers hope that direct neural stimulation could trigger the start and stop of flight. Indeed, Sato was able to show that on and off can be signaled, respectively, by applying repeated positive and negative voltage pulses to electrodes implanted between the left and right optic lobes of the insect’s brain. Once the beetles were in the air, the average flight length was about 45 seconds. Furthermore, Sato found that stringing varying numbers of flight initiation pulses together allowed him to control the vertical tilt of the beetle’s flight and thus control altitude.
To achieve left–right control, they turned to the muscles. An anatomical dissection of the beetle’s flight muscles revealed that a muscle underneath the wing, the basalar muscle, is the most important and easily accessible muscle for controlling the turns. As Maharbiz suspected, direct stimulation of these muscles with low–frequency electric pulses acted as basic turning commands.
Despite these initial successes, the engineers were subject to another key constraint to satisfy the DARPA project goals. Wireless flight requires that the insect carry the weight of the receiver electronics. Beetles can carry up to 30 percent of their own weight, and the team needed a beetle that could carry up at least 1300 milligram of payload weight. Cotinis, however, weigh only one gram, and thus posed unrealistic constraints on what the beetle could carry. Enter Mecynorrhina torquata. Weighing in at just over eight grams, these green iridescent flower beetles, primarily reared as pets for hobbyists, were up to the task.
Each beetle is outfitted by hand one at a time. A bit of beeswax is used to hold the anesthetized beetles in place while Sato pierces small holes in the cuticle on their head and underneath their wings. Steel wire electrodes, already soldered onto the electronic control board, are then threaded through the holes to the appropriate depth in the brain. The board—including a tiny computer running control software, a radio receiver, and an antenna—and a battery are strapped to the back of the beetle like a backpack and the beetle is ready to be flown.
Although ideal for these prototypes, Mecynorrhina torquata* are obviously disadvantaged by their large, bumbling size. Even Maharbiz has admitted that the technology, as it stands right now, would not be effective for stealthy surveillance. “Can you imagine,” he says, “if you were in a meeting and a giant beetle half the size of your palm, completely nonnative to your environment and wearing a circuit board on its back suddenly careened in through the window and crashed onto your computer screen?” Such a scenario would certainly not be very stealthy.
One aspect of this scene could be changed easily. The circuitry could be concealed by implantation in a pupa, which has almost the same anatomical structure as a fully formed adult. Pupal implantation was one of the original contract stipulations, with the idea that neural integration would be more intimate if the neurons were able to grow around the electrode. In reality, the Maharbiz team found no functional advantage and no longer bothers with the technique because of the increased time and effort needed to coax the pupa through metamorphosis. If needed, though, the entire circuit board could be inserted into the pupa, leaving no visible traces of electrical tampering in the metamorphosed adult, whose cyborg status could thus remain undetected.
Full throttle ahead
There are other immediate scientific questions and engineering challenges that the team aims to address in the upcoming year
or two. For one, the control of beetle flight is still shaky, based only on direct user stimulation of the beetle’s flight muscles. To really gain exquisite control of in–flight turning, feedback needs to be implemented directly on the circuit board. “What we want to do is make a synthetic control loop running in tandem with the biological control loop,” Maharbiz says. “I don’t think this is something that’s ever been tried in free flight.”
Maharbiz and Sato, along with an ever–growing group of students, postdocs, and collaborators, are now trying to fully characterize the main muscles that cause turning. That is, given some input, they would like to know exactly what the output is in terms of degrees of rotation. “We want graded behavior,” Maharbiz says, “not just, I hit it and it turns some indeterminate amount.” Using techniques first demonstrated by Michael Dickinson at Caltech, who specializes in fly flight, they aim to vary both the frequency (how quickly the positive pulses are applied to the muscle) and also the phase shift (at what point in the wing beat the pulse is applied) to generate elaborate descriptions of the beetle’s flight response, which can then be used as programming guidelines to implement on–chip feedback.
To monitor the beetle’s flight, the researchers are installing a 12–camera Vicon system, the kind used by animation companies for motion capture, in a large room at UC Berkeley’s Richmond Field
Station. Slap a few one–millimeter reflective stickers on the beetle, throw it into a room outfitted with Vicon cameras, and the system will automatically track the three–dimensional trajectories of the flying beetle. The centimeter–scale resolution provides the necessary precision that is not obtainable using a single camera.
Maharbiz also wants to expand his work beyond beetles. “We aim to soon be able to scale this technology down to where we can actuate a fly,” Maharbiz says. “There’s no reason why this shouldn’t be possible.” Innovations in battery technologies would drastically reduce the payload, of which nearly 30 percent is battery weight. Additionally, the flight control mechanisms of flies are well articulated in the insect literature, lowering the knowledge barrier to robust flight control of these species.
A look to the past
If the cyborg beetles make it to the battleground, this would certainly not be the first time insects have been used for military purposes. “Bees have been used in warfare very explicitly, but in very different ways,” says Kosek, who is working on a book that discusses the recent militarization of bees. As early as medieval times, armies were catapulting raging beehives at oncoming enemy troops or over the ramparts of enemy fortresses.
In the last few decades, military uses for bees have become much more sophisticated. According to Kosek, researchers at the University of Montana began to develop ways to use bees for environmental monitoring and landmine detection. Owing to the electrostatic charge on their branched hairs, bees are essentially aerial feather dusters, collecting all kinds of particles and environmental contaminants as they go. By analyzing the bees’ bodies and contents of the hive, researchers could construct coarse–grained geographical maps of trace chemical contaminants including biological warfare agents, explosives, and landmine vapors. Bees can travel within a three– to five–kilometer radius of the hive, setting the spatial resolution for these passive detection techniques.
This range, however, is too large to be of much use for landmine detection, so the researchers refined their technology with the help of a DARPA contract. Bees can be trained using Pavlovian techniques to become sensitive to particular chemicals, much like dogs are trained tosniff out narcotics. By feeding them trace chemicals mixed with sugar water,researchers can sensitize bees to chemicals such as those present near landmines. Once the bees are thoroughly conditioned, their free–flight patterns can be imaged dynamically across a field using light detection and ranging (LiDAR), a technology that is similar to radar but uses laser light in place of radio waves and can locate the insects hundreds of meters away. In trials, bees were able to successfully locate 10 of 12 planted mines.
With this success, Los Alamos National Labs picked up the work from the Montana researchers and splintered it off into its own independent department, the Stealthy Insect Sensor Program. Rather than using bees to spatially map their environment, this program aimed to use them as lab sensors, using the unfurling of their tongues (proboscii) to quantify the concentration of deadly or toxic chemicals in a given sample. In general, much of this bee research has shifted underground to off–access arenas, even within the national labs. “People I used to interview at Los Alamos, I can’t interview anymore,” Kosek says. “Given this political moment, with the rise of militarism, much of this biotechnology research is going towards creating a new ecology of empire,” Kosek says, contending that the military’s financialinvolvement strongly dictates the direction of the developing technologies.
Up in the air
Kosek intimates that the grossly disproportionate funding of military applications for insects is disquieting. He says that the military has been investing vast sums into bee research, far outstripping governmental investments into their agricultural applications. In the hands of the military, promising technologies can often become classified, making the flow of information asymmetric. In that sense, DARPA–funded university research, such as the work done by Maharbiz and his team, may help to restore some balance to the tipped scale by ensuring that at least some of the work is made available in the public domain. Tiny insect spies may be a worrisome thought, but at least we now know this technology may soon exist.
Even Sato himself believes that many interesting applications of his work will bear fruit in other orchards. “Maybe in the future we can apply this kind of bio–interface to humans.” Eyes widening, he suddenly realizes the implications of this statement. “No,no,not to control humans. I mean, maybe insight gained from this technology could eventually be used to design neural–interfaced prosthetics for people.” Like other researchers developing micro air vehicles, Sato also believes that these cyborg insects may also be useful for locating trapped victims in natural disasters.
Maharbiz refrains from much speculation about future applications altogether. “I want to build these things because they’re just plain cool,” he says, eyes alight with enthusiasm. “I really can’t tell you what it will be used for in the future.”
“The fact of the matter is,” he says, “people have very little predictive power over what a technology might be used for in the future. The Internet is a brilliant example of this.” Developed originally by DARPA in the 1960s and 70s, no one ever predicted that it would be the technological tour de force it’s now perceived to be. There is no question that military–funded research has brought about technologies that have profoundly impacted modern society, including the Internet, the computer mouse, and global positioning systems. These technologies, though invented with military intent, underwent a paradigm shift and were redefined as consumer products.
Likewise, cyborg insects may find other uses as search–and–rescue tools for natural disasters or as distributed sensor networks for environmental monitoring of global warming. It’s still up in the air.
Sisi Chen is a graduate student in bioengineering.
*A mistake in the print edition shows this term as Cotinis instead of Mecynorrhina torquata. We apologize for the mix-up.
The following is an inset box with supplemental background information. It is also found in the print edition of the article.
The military’s purse strings
Close to 70 percent of all engineering research in the United States is funded by the Department of Defense. Although the course of the future may be hard to predict for specific applications, the large coffers of military agencies set much of the overall tone and bias of national research efforts. In 2008, the Department of Defense commanded $6.9 billion to fund science and engineering research, significantly greater than the National Science Foundation’s $4 billion. (The heaviest hitter by far, however, is the Department of Health and Human Services, which has an annual budget of nearly $30 billion, disbursed mainly through the National Institutes of Health.)
Yet the military exerts its influence through more than the heft of its wallet. For instance, DARPA is characterized by creative ‘blue–sky’ research ideas and tight discipline enforced in its review process. Some of its projects are indeed huge flops (think full–size mechanical elephants or Orion, the interplanetary spaceship designed to be propelled forward by launching nuclear bombs out of its rear end), but many of DARPA’s success stories might have never been realized under a more conservative funding scheme. Unlike most grants, which are peer–reviewed, the DARPA contracts are zealously managed by dedicated program managers, who hold awardees to exacting standards and well–defined goals. Contracts are renewed only if stated goals are met. In this sense, other civilian funding agencies might have much to learn from this innovative and results–oriented organization. The new ARPA–E (Advanced Research Projects Agency for Energy), just launched by Energy Secretary Steven Chu last summer, may be a good first step in this direction.