DARPA’s Disruptive Technologies
Nothing quite like it had ever been attempted. Deep in the California desert last March, as a few fatigues-clad U.S. MARINES stood nearby, researchers from the University of California, Berkeley, fiddled with a 1.5-meter airplane with six walnut-sized bundles of electronics ATTACHED to the undersides of its wings. Each bundle, swaddled in pink plastic, held a magnetic-field sensor, short-range radio transmitter, antenna and microprocessor run by a custom low-powered operating system dubbed “Tiny OS.”
And then the remote-controlled plane, freighted with the early embodiments of a hoped-for advance in miniaturized, networked sensing, buzzed aloft, traveled about two kilometers and dropped its pink payload along a dirt road. Soon, as planned, a few trucks drove past the innocuous electronic spies. The bundles detected the trucks’ magnetic fields, shared this information among themselves and beamed a report on the vehicles’ location, speed and direction to the remote-controlled plane circling overhead. The aircraft, in turn, relayed the news to the researchers and soldiers waiting on the rugged terrain of the Marine Corps base in Twentynine Palms, CA.
The bundles were crude prototypes, and it took days to get even this limited experiment right. But someday thousands of similar devices-only much tinier, perhaps as small as dust motes-might be deployed to collect and process a rich array of information about enemy movements, crop conditions, pollution or anything else requiring monitoring. Realizing such a vision will demand advances in everything from microscale sensors to materials to programming. It’s a huge undertaking. But there’s a common benefactor: the U.S. Defense Advanced Research Projects Agency, which brokered the desert experiment and is funding ambitious investigations into each of the technologies involved.
Commonly known as DARPA, this is the U.S. Department of Defense’s storied outpost of technology research-military systems, yes, but also innovations that sometimes create and transform industries. Formed in 1958, in the technological frenzy sparked by the Soviet Union’s launch of its Sputnik satellite, DARPA boasts a four-decade-long history of promoting novel technologies-today doling out nearly $2 billion annually to corporate, government and university researchers in support of high-risk, potentially high-impact ideas. Among its many successes (see “Four Decades of Success,” p. 45), DARPA’s gambles proved instrumental in spawning the Internet and the computer mouse, stealth aircraft and the chip that makes your cell phone work-advances that meant research as out-of-the-box in its time as dust-mote-sized sensors seem today.
DARPA is hardly the only player funding cutting-edge research-think National Science Foundation or National Institutes of Health-and certainly not the deepest pocketed. But the agency’s swashbuckling style of betting on seemingly far-out research-and bringing together interdisciplinary teams that it pushes toward a practical advance-sets it apart. And while some contend that DARPA has moved back from the cutting edge in recent years, concentrating too much on short-term military issues rather than truly breakthrough ideas, no one denies that the agency remains a powerful engine of technological change. “An awful lot of the good stuff we have today is there because DARPA was willing to take a chance on visionary projects,” says David Waltz, president of the NEC Research Institute in Princeton, NJ. “They are the visionary agency.”
In its physical aspect, DARPA is nothing if not mundane. The critical decisions that agency officials make on exotic technologies are rendered in an unremarkable leased office building in Arlington, VA. There are no labs here; DARPA is a funding agency, not a research facility. No sign advertises DARPA’s tenancy to passersby. Except for the anti-eavesdropping gadgets glued to its conference-room windows, this edifice of black-hued glass could pass for an insurance company. But once in the lobby, newcomers must submit their social security numbers to “Visitor Control.” Guards ask, “classified or unclassified?” and make sure guests stay in sight (around here, it’s a no-no to wander into a hallway in search of the water cooler, and telephones bear labels warning that conversations are recorded).
The offices are filled with about 240 employees, of whom half are technical staff-program managers whose job is to shape the work DARPA funds and scour the country for promising new ideas. In keeping with DARPA’s antibureaucratic ethos, these managers are not career government employees but experts on loan from universities, corporations and federal research labs, pulling stints at DARPA of between three and five years. “DARPA is the [Department of Defense’s] center for revolutionary ideas. It is a true bottom-up organization where program managers are the heart and soul,” says Anthony J. Tether, the agency’s director. “We hire people who have a dream that they cannot get fulfilled elsewhere….DARPA program managers are by nature risktakers; they are passionate about making a difference.”
It is in the arena of emerging technologies-funding for research makes up 56 percent of the agency’s $2.2 billion 2002 budget-that the greatest triumphs have come. And a look at the agency’s current lineup shows plenty of potential for future successes. Want microscopically small machines? DARPA was an early funder of efforts to produce miniature mirrors, sensors and gauges-devices used in so-called microelectromechanical systems (MEMS)-that are now widely employed in industry. Want tiny, low-powered computers? DARPA is backing work on logic and memory components as small as individual molecules. Want thousands of sensors (or little robots) to synthesize observations and coordinate actions? DARPA is funding the networking technologies and software they’ll need. Want something to quickly detect tiny amounts of viruses and other pathogens? DARPA is working on that, too, and a lot more.
It all adds up to a diverse panoply of projects, but the principle on which they are chosen is the same: “We’re about surprise. Prevent surprise, and create surprise,” says Jane Alexander, the agency’s deputy director. “You need a skunk works, somebody over in the corner who is anticipating what your opponent is doing and what you are going to answer that with, and also is anticipating what your next generation is-what are you going to surprise somebody with. DARPA is that thing.”
But the key to the agency’s success lies not so much in its mission as in its unique administrative model and management philosophy. For starters, before DARPA officials even decide what to fund, “one of the questions senior management asks is, Is somebody else able to do this problem?’ If they are, let them do it,” says Alexander. And if not, DARPA primes the pump-providing enough time and money for the technology to take root in the commercial world. “With the right investment at the right time, I can steer industry toward an area that will be useful. I nudge them.”
These are multimillion-dollar nudges, of course, so the aim is to choose carefully. After hearing from the military about its near- and long-term needs, DARPA’s program managers design multidisciplinary programs to help meet them. Major initiatives-or “thrusts”-usually last four years and incorporate five to 10 research teams; funding typically runs between $10 million and $40 million, and occasionally much more. Whatever the scale, though, DARPA stresses teamwork among research groups and enforces short-term performance milestones. And then, just as feverishly as DARPA begins a thrust, it often pulls out. Either the teams are unable to meet their goals, or they succeed sufficiently that the commercial sector or other research- funding sources pick up the ball. Alex Roland, a Duke University professor of military and technology history, says 85 percent of the agency’s programs fail. “It’s not an aspect of what they do that they want publicly displayed.” Roland says. However, he adds, “You’ve got to expect a high rate of failure because the payoffs are fabulous.”
So where will the next big successes come from? TR canvassed DARPA directors to identify today’s hottest research projects. There’s no guarantee any will pan out. But together they provide a representative look at the agency’s most cutting-edge initiatives-and the direction technology is heading.
Two seemingly unrelated events in early-1990s materials research have evolved into one of DARPA’s most intriguing new areas of focus. One was the air force’s ongoing quest for stronger, lighter materials to build better planes. The second exploded into view after the Persian Gulf War. During the conflict, United States-led forces used shells made of radioactive uranium-238 to attack Iraqi tanks. Instead of flattening on impact, uranium-238 peels away in layers and actually sharpens, making it more destructive than conventional shells. But some veterans’ groups soon claimed the radioactive residue caused health problems. Plus there was an expensive environmental cleanup required. All this led the army to seek a nonradioactive replacement for its uranium projectiles.
In the mid-1990s, these parallel needs led DARPA to the Caltech lab of William L. Johnson, a pioneer in a field known as “glassy metals.” Such materials look like ordinary metals, but they have a key difference: they’ve been fabricated so their atomic structures aren’t orderly, or “crystalline,” but rather random or “amorphous” in nature-like the atomic structure of glass. Scientists have known for at least a decade that a random atomic structure in a metal alloy can confer greater strength and more resistance to fracture and corrosion than are provided by crystalline structures, which contain more defects that make for weakness than amorphous structures. The problem is, glassy metals are extremely difficult and expensive to produce. In most cases, therefore, they have only existed as laboratory curiosities (an exception is Johnson’s glassy zirconium-beryllium alloy, now used in high-end golf clubs). But working under army sponsorship, Johnson’s lab in 1997 came up with a glassy tungsten that not only self-sharpened-making it a potential replacement for uranium shells-but pointed the way to techniques for mass-producing glassy metals with broader applications.
Looking to direct more firepower into this potential breakthrough area, DARPA this spring began a four-year, $30 million thrust to fund efforts to model the atomic interactions that take place as metals are mixed and cooled. The hope is that this insight will lead to glassy versions of widely used metals such as aluminum, titanium and iron that can be fabricated by the ton in existing factories. The first glassy metals were discovered “by trial and error, by happenstance, some might even say alchemy,” says Leo Christodoulou, DARPA’s manager of the new program, called Structural Amorphous Metals. “What we are trying to do is put [more] science behind this program, try to understand the fundamental physics.”
DARPA’s effort attacks the problem from several different angles. For starters, a team led by Johnson that includes seven university labs and three military research groups will do the underlying scientific studies and computational work and create new samples. The prototype materials will then be passed to industrial partners for small-scale fabrication and testing. (As of mid-August, the partner companies had not been announced.)
Whether any fundamentally new, factory-ready-metals recipes will emerge from this collaboration is an open question. But the potential payoff is clear: Johnson’s group, for instance, is working on glassy aluminum and magnesium alloys that would possess twice the strength of their crystalline counterparts. That means less material would be needed to, say, build a fighter jet or a 747, enabling it to save fuel or carry heavier payloads. “If we can successfully do this, then this is the material aircraft will be built out of in 15 years,” Johnson says. “It will become a major paradigm shift in the way we use metals.”
The fusion of computing with other fields has become a given in recent years. The combination of computers and communications forms the basis of the Internet. The application of computing power to drug development has spurred bioinformatics and other, related areas of genomics and proteomics. In DARPA’s view, the next challenge will be linking biology and computing to the science of the very small, through devices that can detect, influence, interpret and communicate what’s happening in living cells. And so DARPA this year kicked off an ambitious $35 million, four-year effort called Bio:Info:Micro. As Alexander told a group of researchers last fall, there’s a growing sense that merging biology with computing and microsystems “is something really new and revolutionary. In a lot of cases, we can’t quite put our finger on it, but all of us, as technologists, think that this is a very promising area.”
Two basic programs aim to fire early salvos in this predicted revolution. The first attempts to advance brain-machine interfaces-technologies that tap brain signals to control a variety of mechanical and electrical devices and can also send signals into the brain to stimulate neurons. This program has a solid starting point: already, DARPA-funded groups from Duke University, Caltech and elsewhere have built devices (tested only on animals so far) that can be surgically implanted in the brain to detect neural signals and send those impulses via wires to computers. The computers decode the signals, then transmit control instructions to devices like robotic arms (see “Brain-Machine Interface,” TR January/February 2001).
Geoffrey Hinton tells us why he’s now scared of the tech he helped build
“I have suddenly switched my views on whether these things are going to be more intelligent than us.”
Meet the people who use Notion to plan their whole lives
The workplace tool’s appeal extends far beyond organizing work projects. Many users find it’s just as useful for managing their free time.
Learning to code isn’t enough
Historically, learn-to-code efforts have provided opportunities for the few, but new efforts are aiming to be inclusive.
Deep learning pioneer Geoffrey Hinton has quit Google
Hinton will be speaking at EmTech Digital on Wednesday.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.