Small wonders

Smart dust, robotic flies, and civil liberties. Published in California Monthly magazine.

When Kris Pister first came up with the name “smart dust” to describe the tiny, wireless, electronic sensors he began designing in the early ’90s, he says he meant it as a joke. “Everybody was talking about smart houses and smart bombs and smart this and that, so I figured I’d give the world ‘smart dust,’” recalls the professor of electrical engineering. The name stuck, but there’s nothing frivolous about smart dust itself—future uses are said to include covert surveillance systems that James Bond would marvel at; heat, light, and movement sensors in buildings, which could save billions of dollars in wasted energy; and virtual keyboards able to detect the movement of fingers in thin air.

Individual smart-dust units, or “motes,” contain a sensor, a tiny computer, wireless communication capabilities, and their own source of power. Yet each mote could eventually be no bigger than a grain of sand, or even smaller. Pister first got the idea for smart dust in 1992, when it was clear that both computers and wireless communications devices were shrinking rapidly and would continue to get smaller. He predicted that sensors built with the then-emerging technology called MEMS—micro-electro-mechanical systems—would shrink at the same extraordinary pace. Pister’s “smart” idea was to combine all three shrinking technologies in one very tiny, but very powerful, unit. “It’s completely autonomous,” he explains. “It lives all by itself and can talk to other things.”

There is a MEMS sensor to detect just about anything—heat, light, movement, magnetic fields, chemicals, even biological molecules. Like computer chips, MEMS devices are made by etching silicon wafers, and thousands of sensors can be etched onto a single wafer, making mass-production possible. But while microchips only harness silicon’s electrical properties, MEMS devices also make use of silicon’s mechanical properties. These tiny machines have real moving parts—sometimes miniature gears, springs, or levers—that can switch electronic circuits on or off. While silicon may be nearing its electronic limits, its mechanical properties are just beginning to be explored. So, while some predict the end of silicon, and are pushing for nanotechnology to move in to take its place, those who work with MEMS—including dozens of researchers at Berkeley’s Sensor and Actuator Center (BSAC)—say that reports of the impending demise of silicon have been greatly exaggerated.

While most of us have been unaware of the technology until now, many of us already rely on it—for example, a MEMS sensor is what tells the airbag in your car to deploy after a collision. There may even be biomedical applications in MEMS’s future. Professor of mechanical engineering Arun Majumdar has developed a device that uses microscopic cantilevers for the early detection of prostate cancer. Cancer-related proteins stick to antibodies on the cantilevers, bending them ever so slightly.

Pister has shown off the capabilities of smart dust to the U.S. Department of Defense (DOD), which funds his project through its long-term research arm, the Defense Advanced Research Projects Agency (DARPA). As a demonstration, he sent out a small remote-controlled plane over a Marine Corps base, where it dropped some prototype smart-dust motes containing magnetic sensors. Using computer science professor David Culler’s specially designed computer operating system—called TinyOS—the motes instantly formed a network, detected different vehicles passing by according to their magnetic signatures, and relayed that information back to the plane. “The application for [the military] was very clear—to know where all the bad guys are,” says Pister. Other types of smart dust could allow soldiers to detect chemical or biological agents on the battlefield before the soldiers get there themselves.

The DOD funds much of the work in MEMS, around the country and at Berkeley, where the director of BSAC, Albert Pisano, is the former manager of DARPA’s own MEMS program. Some people are uncomfortable about the high level of military research carried out in American universities. In the federal budget proposed for 2002, military programs account for 69 percent of all research done in the physical sciences and engineering, according to figures obtained by Charles Schwartz, emeritus professor of physics. He says that the Star Wars program of the ’80s generated some debate on campus about the relationship between the military and the universities, but that public discussion soon petered out. He would like to see the debate revived: “The role of the university should be continually questioning those type of things, not making it easy for people to fall into ruts,” says Schwartz.

Engineers point out that it’s not just the military that benefits from their research. “Virtually any technology that we develop under DOD funding has immediate commercial applications,” asserts Pister, who already has interest from the campus in using his vehicle-sensing smart dust to monitor traffic flow. Another exciting possibility is a virtual keyboard—a smart-dust mote stuck to each fingernail could allow finger movements in air to be transmitted to a computer. With this technology, computers could get even smaller, and air guitar would no longer be just a fantasy of rock-star wannabes. One of the most widely touted commercial applications is the use of smart-energy sensors in office buildings, which typically waste huge amounts of energy as computers, heating, and lighting remain switched on in empty rooms. Smart-dust sensors costing just a few dollars each could be stuck to the walls of a room—even painted on—to detect whether someone’s there, and to monitor room conditions. That data would be fed back to the building’s heating and lighting systems, which would then adjust accordingly. It has been predicted that smart sensors could ultimately save around $7 billion in wasted energy each year in California alone.

Those sensors might one day save more than just money, says Pister: “A complete temperature map of the two [World Trade Center] towers would have saved lives in the evacuation of those buildings—there were certainly people who could have gotten out if they had only known where to go to get past the fire.” He adds that the sensor network would have survived the buildings’ collapse, and could have detected heat or sound to guide rescuers to survivors.

The immediate goal of the smart-dust project is to build a mote that occupies just one cubic millimeter. Their latest model, called golem dust, is tantalizingly close, at just 4 cubic millimeters. The name came from graduate student Bret Warneke, who explains that golem is a Yiddish word that roughly translates as “automaton.” The name has turned out to be more fitting than Warneke realized—in Jewish mythology, a golem is a creature with no soul but with the capacity for either good or evil—something that has also been said of smart dust. When smart dust is used for surveillance, people have no way of knowing they’re being watched because the sensors are so small. While this means that government agencies like the CIA, FBI, or National Security Administration could use the technology to monitor the activities of suspected terrorists, civil liberties groups are concerned that those agencies might also be tempted to use it to monitor citizens in a more sinister way.

Pister is exasperated by media reports that focus on the “Big Brother” applications of smart dust: “I’m totally not worried about the government using this—I’m much more worried about my fellow man,” he says, meaning that smart dust would make possible incredibly cheap, and totally inconspicuous, surveillance of anyone by anyone. “I think it’s critical that we in the universities warn people well in advance [about these privacy issues] so that the response is rational,” he continues. Smart-dust engineers have already been consulting with scholars at Berkeley’s Center for Law and Technology on how best to address such concerns.

Law professor Deirdre Mulligan believes that current privacy laws certainly need to be updated, but warns that laws alone may not be enough: “In a world covered with smart dust, there’d be pictures of you every place you were, every day, every moment, whenever you’re out in a public space. I don’t know that there would be a legislative response to that that would be satisfactory.” But Mulligan is heartened that Pister’s group is contemplating potential misuses of smart dust now, well ahead of any commercial applications. “At this point, it’s important to think of how you can build in protections to the technology itself, because once it’s out there, it’s very difficult to put the cat back in the bag. I’m certain that there are design decisions that can be made early on that will favor privacy,” she notes.

Worries about the adverse aspects of smart dust may have given Pister pause, but they have not held him back. A science-fiction addict, he has long fantasized about building a synthetic insect—“smart dust with legs,” as he likes to call it—and is well on his way to realizing that dream. So when he heard of Arun Majumdar’s even more far-out idea—to build a synthetic flying insect—he jumped at the chance to be involved. The Mechanical Flying Insect (MFI) project, as it is called, is also funded by DARPA. Under the direction of electrical engineering professor Ron Fearing, it is a collaboration between a number of groups across campus to build what is essentially “smart dust with wings,” although that description doesn’t begin to capture the technical challenges of the venture. “What we thought conceptually would be a very simple thing is conceptually a very difficult thing,” says Fearing. “It’s been a lot of hard work and a lot of finding things out the hard way. Everything along the way—from the joints to the structure—had to be designed completely from scratch.”

For one thing, modeling the rapidly beating wings of an insect turns out to be exceedingly complex. Using rudimentary principles of aerodynamics, an engineer back in the 1930s famously “proved” that a bumblebee shouldn’t be able to fly. And at the start of the MFI project, in 1998, scientists still had no idea how flies actually fly. Luckily, Berkeley’s resident fly expert, professor of integrative biology Mike Dickinson, had been working on the problem for almost a decade, and a few months into the project he finally came up with the first real explanation of how flies stay aloft. Since then he has determined all of the basic principles of insect flight, and has passed that knowledge on to the engineers, who are trying to copy what flies do—in just 100 milligrams of steel and plastic.

To get to the heart of the problem, Dickinson built a giant robotic fly down in the basement of the Valley Life Sciences Building. Nicknamed “Robofly,” it consists of two large plastic wings that sit in a giant vat of mineral oil and move just like a fly’s wings—simulating the forces and fluid flow around the wings of a real fly. Robofly showed that the rapid beating of a fly’s wings, and the quick rotations they perform at the end of each beat, together generate enough lift to keep the fly in the air. Dickinson has also built a “virtual reality arena,” complete with smell-o-vision, to study real flies and get an idea of how they use their sensory systems to execute those extraordinary maneuvers that make them so difficult to swat. Fearing plans to copy some of those strategies in the smart-dust “brains” of his mechanical fly.

Despite the hurdles, the MFI group has so far managed to build a lightweight, mechanical, fly-on-a-stick model with operating wings that beat 200 times a second. They hope to have a complete, working robotic fly by the end of 2003. At that point, they could make a whole swarm of robotic flies. Pister jokes that he would take his swarm to barbecues and use it to enforce his own personal “no-fly zone.” More serious applications include using swarms of robotic flies for military surveillance and intelligence gathering.

Dickinson also envisages that swarms of mechanical flies would aid in search-and-rescue operations, for example in the rubble of collapsed buildings. They could be designed to detect the small amounts of carbon dioxide coming from the breath of survivors. “One of the problems in search and rescue is that you have to find people really quickly. If rescuers knew where the survivors were, they’d know where to dig. Hopping through from space to space, these [tiny robotic flies] could get through to crevices very quickly,” he predicts.

The potential of this technology—for both good and bad—causes Pister to be reflective: “I have thought about this a lot, and I would not be doing this if I thought that it was going to be a negative thing for humanity,” he concludes. But Schwartz would like to see continuing engagement with such issues by the campus as a whole. “Scientists have an obligation to think about what their work might be used for, but I would like to see that become a public debate, and not just a private debate the scientist has with himself,” he says. “Who knows—we may all agree in the end that this technology does more good than harm. But these are public issues, so there should be public discourse and debate.”

The next small thing

You’re looking at the world’s smallest transistor (where the blue lines cross, just a few atoms thick). The lines are carbon nanotubes, cylindrical molecules of strongly bonded carbon, which have a strength more than five times that of steel and almost no electrical resistance. The transistor was built by physics professor Alex Zettl, who has also used nested carbon nanotubes as tiny bearings that are wear-free and almost completely frictionless, and which could be used to lubricate the tiny gears used in some micro-electro-mechanical systems (MEMS). Zettl says that in the future they might even be used for NEMS—nano-electro-mechanical systems—which would be around the size of molecules, a thousand times smaller than the already tiny MEMS devices. Zettl is one of nearly 100 scientists at Berkeley involved in nanotechnology-related research. The field got a huge boost at the beginning of last year, when President Clinton announced his National Nanotechnology Initiative, allocating hundreds of millions of dollars of federal funds to research in the area. The initiative was in part the idea of Tom Kalil, then a science advisor to President Clinton, and now science advisor to Chancellor Berdahl. “We’ve been on this Moore’s Law curve of being able to double the capacity to store and process information every 18 months. But the technology we’ve been using to do that may run out of gas at some point,” says Kalil, who predicts that electronic circuits made of carbon nanotubes will move in to revolutionize computing, one day making it possible to store the entire Library of Congress in a device the size of a sugar cube.

Honey, I shrunk the engine!

Engineering professor Carlos Fernandez-Pello proudly shows off his newest creation—the world’s smallest combustion engine. While micro-electro-mechanical systems (MEMS) have shrunk at an incredible pace, the batteries that power them have stubbornly refused to follow. “Batteries are heavy, they don’t last too long, and they are an environmental hazard,” says Fernandez-Pello. “But hydrocarbon fuels have an energy density around a hundred times higher than a battery.” This mini-engine is no bigger than a penny and runs on lighter fluid. Fernandez-Pello is now working on a 3mm engine and a 1mm engine, both made of silicon. There have been some unexpected early payoffs from his research. “To me the mini-engine was just a research tool, but now there is a lot of commercial interest in it,” he says. The mini-engine could be used to produce as much as 30 watts of energy, enough to power a whole range of household devices, including laptop computers and cell phones.