Chahab Nastar has always had a passion for patterns, whether he was tracking the popularity of his rock band Busy Being Born or designing software to distinguish unhealthy heart motions, which he did as a researcher at the University of Paris. Now he’s pouring that passion into helping computers understand everyday objects and scenes—a challenge that’s still “extremely difficult, if not impossible, for machines,” he says. Show a picture of your beach vacation to today’s average image-recognition program, and it can find similar beach scenes. But it can’t tell which beach is pictured or who is sunbathing. To move to that level, Nastar’s Paris-based startup, LookThatUp—now LTU Technologies—has enhanced image-recognition techniques with artificial-intelligence- based learning algorithms. Now, the more images of, say, cars and animals LTU’s system memorizes, the more quickly it can classify a Volkswagen or zebra. The system’s industry-leading recognition speed, a mere 200 milliseconds per image, has helped LTU sell software licenses to a dozen U.S. and European companies.
No pictures adorn Christopher Ahlberg’s office. The Swedish-born computer scientist and amateur kickboxer says simplicity encourages people to do what’s needed and move on.This philosophy pervades his company,Spotfire.Its DecisionSite software allows computer users to go into multiple databases regardless of their format,easily import,export and manipulate very large data sets,and visualize the results in dynamic graphs,charts and plots.In other words:get the information they need and get out.More than 400 biotech,pharmaceutical,oil and gas,chemical,and auto companies—from Merck to Saab—now use DecisionSite to help them render verdicts on everything from scientific experiments to manufacturing lines. Founded in 1996 by three people,the privately held company now employs 175 in Somerville, MA,and Go¨teborg,Sweden.Ahlberg has grander designs for DecisionSite,too.Chief among them:becoming the one application companies use to interrogate databases around the world over the Internet,as simply as they navigate text documents on the Web.
For years, researchers have dreamed of improving air traffic safety and efficiency by giving pilots a real-time 3-D display that shows how to navigate terrain, even in bad weather. But merging Global Positioning System data with graphical displays of the earth’s surface proved dauntingly expensive. As a PhD candidate in aeronautical engineering at Stanford University, Andrew Barrows delivered the first practical, inexpensive “highway in the sky. ”He did it by writing software that merges GPS location information with images from terrain databases and shows the pilot a series of rectangles. The pilot need only keep the plane flying through these targets. “Fifteen years from now, every airplane will have a sophisticated version of this, ”says John Hansman, MIT professor of aeronautics and astronautics. While the Federal Aviation Administration works on certifying this type of system, Barrows has left academe to become president of Nav3D in Palo Alto, CA.One big client, Boeing, is adapting his technology for military use.Nav3D is also exploring displays that would help construction crews “see” underground gas lines or guide firefighters through smoky buildings.
Elisabeth M. Belding-Royer
Today’s mobile data networks are spotty. If you’re not within range of a transmitter or are cut off by large obstacles like skyscrapers, you’re out of luck. The solution could be networks that form only when needed, and Elizabeth M. Belding-Royer may deliver them. As a graduate student at the University of California, Santa Barbara, Belding-Royer worked with Nokia research fellow Charles Perkins to develop the necessary network protocols—the operational instructions. If your handheld device finds itself with a nonexistent or failing signal, it can use Belding-Royer’s protocols to find and connect with nearby wireless devices. These neighbors then find a path through still other wireless devices to create an ad hoc but solid connection. Designing the protocols helped Belding-Royer land a professorship at UC Santa Barbara, and the Internet Engineering Task Force is now considering turning them into standards. Applied,the protocols could eliminate “dead zones” that wireless transmitters don’t reach and make it cheaper and easier to set up networks everywhere—from the Sahara to downtown Los Angeles.
Vincent Berger has two jobs, two labs, even twin babies. As a researcher at Paris-based aerospace giant Thales, he developed the technology behind a new short-wavelength night-vision camera for military surveillance. But he’s best known for his theoretical contributions in optical semiconductors, quickly becoming a linchpin of telecommunications. At 29,Berger was the first to describe ways to integrate photonic crystals with bulky optical devices such as routers. Many think this work will lead to wafer-thin chips that can manipulate photons the way semiconductors control electrons. He was also first to demonstrate that light waves could change color in gallium arsenide, the superfast material of choice for the ubiquitous semiconductor laser. While others convert Berger’s theories into technologies such as miniature cryptography devices, telecom traffic busters and air pollution detectors, he is busy balancing his position at Thales with his new role as university professor. He looks forward to forging industry-university partnerships—not often found in France—to boost the commercialization of photonics research.
When most people speak of advanced materials, they mean semiconductors or biopolymers. But Daniel Branagan means steel. The materials scientist is transforming metals industries with his new “superhard steel.” His novel process rapidly cools molten steel into a glasslike solid, then heats the solid to form a unique nano structure. The result is a coating that outperforms the hardest metal coatings, even tungsten carbide, at one-third the price. Superhard steel is cheap and easy enough to manufacture that applications could range from rock crushers to kitchen knives. More than 500 companies have expressed interest in using Branagan’s material to make tougher, lighter, longer-lasting products. Branagan may license the patent from the U.S. Department of Energy—which oversees his lab—to create a spinoff company. The Michigan native launched his career 12 years ago by developing tougher rare-earth magnets for computer hard drives. But while other metallurgists tout Branagan’s innovations as feats of nanoscale engineering, to this ranch owner outdoorsman they’re “just fancy steels.”
Lasers are found in everyday products, from compact-disc players to bar code readers. But in a laser beam strong and focused enough to shoot down an enemy aircraft or chisel pin-sized mechanical parts out of metal blocks, the hyperexcited photons have to be controlled so they don’t scatter. Improvements on solid- state lasers have been stymied by the tendency of a beam to distort as the crystal at the heart of a laser heats up. Arnaud Brignon, who was born eight years after the laser was invented in 1960,has solved this problem by developing a self-correcting mirror made from nonlinear crystals that cancels the distortions. His division of multinational aerospace giant Thales, in Orsay, France, is identifying markets for commercial versions of the laser, which Brignon says could be ready in three years. While working on a next-generation laser, Brignon finds time to hunt for dinosaur remains in the fields of France. He recently discovered a tooth from a 100-million-year-old armored ankylosaur that was previously undocumented in his country.
Hydrogen fuel cells promise to break the world’s fossil fuel habit without a puff of carbon dioxide, and Joseph Cargnelli is helping them deliver. In 1995,from a small room above his family’s machine shop in Toronto, Ontario, the mechanical engineer and two associates launched Hydrogenics. The company made its mark producing test stations Cargnelli designed and assembled to put fuel cells through their paces. The test units accelerated the work of fuel cell developers and secured Hydrogenics’$84 million initial public offering in 2000.Today Cargnelli, vice president, is worth millions. But the roar of equipment still fills his shop floor office at a complex on Toronto’s west side, where Hydrogenics is developing its own fuel cell engines. Last year,in a six-month span, Cargnelli’s team prototyped a fuel cell generator and transformed it into a backup power supply to keep cell tower antennas and their networks alive during blackouts. This work clinched a partnership with General Motors. To Cargnelli, success just means one more step toward the hydrogen economy.
Howie Choset built himself a “snakebot” named Schmoopie, but that’s not what sets him apart from colleagues. The mechanical engineer and roboticist has developed motion-planning algorithms that ensure his autonomous, multi-jointed snakes not only sense and respond to objects in their path but explore every nook and cranny as they traverse a terrain. Other path-planning algorithms leave room for ambiguity, but Choset’s provides for complete coverage. As a result, the U.S. Office of Naval Research is funding Choset to build robots that search for buried mines. Choset has equipped his bots with mine detectors; when they sense a mine, they map its location, then maneuver around it. Choset is also working with Ford Motor to develop robotic car-painting techniques that will save production time. During it all, he has developed the robotics minor at Carnegie Mellon University and says, “I would like to see this work get to the high-school and junior-high level and have students build robots to learn basic math and physics.” Choset also hopes his algorithms and theories will one day be used in non- robotic applications, such as predicting crime patterns.
So much data, so little bandwidth. That’s the mounting problem Caltech professor of electrical engineering Michelle Effros has wrestled with for 10 years. Her innovations in data compression over networks are so original that Stanford University electrical engineering vice chair Robert Gray calls them “profound.” Her heady work has almost single-handedly created research interest in algorithms to optimize transmission of data over busy,noisy networks like the Internet and various wireless infrastructures. At first “it was hard even to get peer reviews of research,” she recalls. But Effros persisted. “I like working further out. Entrepreneurism is not what excites me—research into new areas does.” Today, however, her algorithms set the standard for academic- and commercial-network compression techniques. And the recreational mountain hiker says with a smile that her once neglected field now teems with breakthroughs. “We are really starting to crack the data compression problem,” Effros says. “Progress is coming much faster now.”
Train buff Alexander Fay jumped at his PhD advisor’s suggestion to design knowledge-based software to help human railway-traffic controllers in their struggle to keep trains running on time. When train breakdowns occur, some controllers tinker with track signals to reroute trains; others call for backup trains. Their decisions ripple through the rail system. Studies estimate $200 in lost economic opportunities for each minute a passenger train is late. Given Europe’s congested rail networks, the potential savings from better management is huge. After interviewing dozens of controllers, Fay used fuzzy-logic principles to integrate 150 when-this-occurs-do-that rules into his software. German Railways is now implementing his computer program, which standardizes the most efficient responses to service disruptions, in one of its regional control centers. Fay’s methods also apply to other systems. At Zürich, Switzerland, construction giant Asea Brown Boveri, he is developing software that captures the know-how of process and control engineers to reduce problems in designing and running pharmaceutical, chemical and energy plants.
Beam mid-infrared light waves across a power plant’s smokestack and you can measure the flow of gaseous pollutants. Feed that data back to the plant and you can trim pollution and fuel consumption. It’s just one opportunity created by the quantum cascade laser. The laser, invented at Bell Labs in 1994, had promise as the heart of a smaller, less expensive, more efficient apparatus for monitoring smokestack emissions. But when Austrian physicist Claire Gmachl arrived at Bell Labs in 1996,the laser still had one fatal flaw: a messy, broad-spectrum beam. Within a year, Gmachl had the problem licked. She amplified one portion of the beam by sculpting the laser crystal into an echo chamber for photons. It was a major leap, and colleagues say the intense Gmachl has since delivered an average of two advances of similar importance each year. She’s also been experimenting with lasers tuned to identify telltale gases found in our breath that may indicate everything from asthma to heart disease. Now Gmachl is focused on carrying data in fiber optics, where rapid-pulse lasers could help sate our hunger for bandwidth.
As a child,Michael Hansen hung out at Radio Shack and wrote such good programs on the store’s computers that the salespeople ran them as demos. Software mastered, he learned hardware, earning a graduate degree in electrical engineering. In 1993 he joined Princeton, NJ-based Sarnoff to tackle visual processing— “the hardest darn problem I’d ever seen.” Before he knew it, Hansen, who also found time to become a private pilot, was leading a $5 million-a-year group. In 2000 his team developed a chip that lets inexpensive portable devices process visual data collected by surveillance cameras. The chip provides hundreds of times more visual processing than a general-purpose Pentium microprocessor at one-tenth the cost, says Peter Burt, director of the vision technologies lab at Sarnoff. The for-profit R&D company believes networks of such simple devices will have great commercial value in military surveillance, law enforcement and auto safety. In order to, as he puts it, “shorten the path from technology development to new products,” Hansen is now working on his MBA.
Tom Hyongsok Soh
When Tom Soh was a Stanford University graduate student, his focus was scholarly: to push the envelope in nanotechnology research. Things have changed since he arrived at Lucent Technologies’ Bell Labs. Soh heads optical microelectromechanical development at Lucent spinoff Agere Systems, in Allentown, PA.His task is to make optical communications systems more efficient and intelligent. His group’s first success was a microelectromechanical switching device that routes fiber-optic signals without converting them to electronic form and then back to optical. Those conversions cause the biggest bottlenecks in today’s telecom systems, and Soh’s optical switch offers huge gains in speed and capacity. Agere began large-scale production in March 2002,after it received significant orders from top customers. Soh has since led development of another product, the optical add/drop multiplexer, which allows light-wave transmissions to be added or dropped at critical nodes without electronic conversion. Currently overseeing 12 engineers, Soh believes great things can be achieved with team chemistry.
Matt Keyser revels in pedaling his bike amidst a pack of racers pounding down a precipitous mountain road. During sane moments, Keyser designs transportation technology at the National Renewable Energy Laboratory in Golden, CO. He’s received two patents since 1992,with three more in the works. In 2001,Keyser and coworkers significantly extended the life span of lead-acid batteries in cars. Batteries provide a power surge to a vehicle’s starter, then recharge. Normally, the process encourages sulfuric acid to oxidize, which ruins the negative battery plate. But Keyser changed how the battery charges, reducing chemical reactions and extending battery life up to 400 percent. “That means fewer batteries in the landfill,” he says. Ford Motor is testing the battery in a prototype electric vehicle. In 1997,Keyser wrapped a standard catalytic converter with a vacuum insulator to keep it hot for hours. The retrofit eliminated 80 percent of the tailpipe emissions that a car produces while warming up. The unit is being commercially developed by auto parts supplier Benteler. Keyser particularly likes its adaptability: “You can put it on a new car, or a ’78 Pontiac.”
In Kara Kockelman’s transportation models, there is no place to hide. The civil engineering professor crunches data on where people shop, work and go to school, what kinds of vehicles they buy, real-estate development trends and demographics to help devise optimal transportation policies. Her results are sometimes provocative: for example, from traffic data, she calculated that a large SUV slows traffic by spending as much time lumbering through an intersection as 1.41 passenger cars. Kockelman says, “We can’t build our way out of congestion,” and argues that road users should therefore shoulder the costs they impose on others. Her studies support, for example, the selective enactment of something called “credit-based congestion pricing”: drivers would be allotted a certain number of commuting trips, which they could use or trade, the way power plants trade emissions credits. Such ideas are not particularly popular, but, Kockelman maintains, “I divorce myself from the emotions of the issue and allow the data to tell me what’s happening.” And Kockelman practices what her studies preach: she takes the bus to work.
When he left the University of Michigan, Paul Krajewski was an expert in “creep” behavior related to all things aluminum. In the jargon of metallurgy, creep is heat- and stress-related deformation, and it’s part of the reason aluminum is tricky to use for making cars. Aluminum is about 67 percent lighter than steel, but it is far more susceptible to cracking when deformed. Today Krajewski is a leading materials scientist at General Motors, devising ways to engineer aluminum so it can be used in mass production of lighter and more fuel-efficient family sedans and SUVs instead of just pricey handmade exotic cars. Krajewski invented a flash-heating technique that allows aluminum to bend on the assembly line without cracking. To ease one manufacturing process, he even developed a lubricant using milk of magnesia. Krajewski has won seven patents related to his techniques, which will help General Motors build cars with “creases and curves, anything that would excite the eyes of the customer,” he says. And those lightweight parts should also excite customers when they get to the gas pump.
Christina Lampe-Onnerud’s energy is boundless. She’s a cellist, a master of jazz dance and has directed award-winning choruses. Boundless energy is also her technological goal. During doctoral work in inorganic chemistry in her native Sweden, Lampe-Onnerud patented a new cathode material that increased the power of lithium batteries. After leading research devising high-energy materials at two startups, she joined New Jersey-based Bell Communications Research in 1995.There she helped develop prototypes of the first lithium batteries made from thin-film polymers and the first practical process for manufacturing them. The batteries were smaller, more powerful and safer than conventional lithium batteries. Today the technology is licensed by many major battery makers. With seven more patents filed, Lampe-Onnerud now oversees 10 labs that investigate new battery materials for Tiax, which bought them from Cambridge, MA-based consulting firm Arthur D. Little. As a consultant, Lampe-Onnerud has led research teams at numerous companies. Her goal: “to cram as much power as possible into a battery without it blowing up.”
Singapore leads the world in fighting traffic with technology. Vehicle sensors are ubiquitous, and so are message signs warning drivers of upcoming jams. Drivers are even charged higher tolls during rush hour. But when the island government seeks to boost the IQ of these “intelligent highways,” it turns to Der-Horng Lee, a civil engineer at the National University of Singapore. He is in demand among transportation engineers and companies worldwide, helping them write better software that models traffic and controls road signals and signs in real time—for example, changing urban stoplight sequences on the fly as drivers flee rush-hour highways. Lee says traffic prediction is like weather prediction, only tougher. Predictions must be done quickly and account for the fact drivers might change plans after hearing them. The key is having the right algorithm, based on the right traffic simulation model. Most models assume group behavior, but Lee’s “microscopic” models acknowledge that some traffic-bound drivers will sit in the mess while others will cut through city streets or choose a longer highway route.
Most matter isn’t very smart. But some exotic materials have memory: you can bend them, but when heated they return to their original shape. Engineers have tinkered with these materials for robotic and automotive applications, but polymer chemist Andreas Lendlein envisions their use in implantable therapeutic devices. In 1997, while working at MIT, Lendlein became the first to develop a biodegradable shape-memory polymer that responds to body temperature. A surgeon could insert a compressed polymer through a tiny incision; once inside the body it would expand. The payoff could be improved coronary stents to prop open blocked arteries, or scaffolds for growing new organs. A polymer with cells attached could be inserted to replace lost cartilage; triggered by the body’s warmth, the polymer would expand into the shape of the missing cartilage, then degrade as new tissue grew. Lendlein returned to his native Germany and cofounded mnemoScience, in Aachen, to commercialize his technology. MnemoScience’s researchers have successfully tested Lendlein’s materials in animals, and they hope to release their first medical product in a few years.
Ihor Lys wants to color your world with light that morphs, fades and blends in computerized patterns, thanks tomultihued arrays of lightemitting diodes (LEDs). Lys came to lighting after 11 years at Carnegie Mellon University, where he studied electrical engineering and robotics. While working on an LED-based display for a robot, he realized that by combining bright blue LEDs—just then being reported in labs—with existing red and green ones, he could create new possibilities for digitally controlled illumination. But it took circuit design virtuosity to produce lush visual environments using these simple indicator lights. In 1997 he teamed up with engineer George Mueller and launched Boston-based Color Kinetics, which reported revenues of $17 million in 2001.Colors from its LED fixtures fill corporate lobbies, swimming pools, spas—and even emanate from the cables of Philadelphia’s Ben Franklin Bridge. Lys “keeps pulling off miracles,” says the MIT Media Lab’s Michael Hawley, a Color Kinetics board member. But Lys views his mission in simple terms: “I see things that are expensive and difficult, and I want to make them cheap and easy.”
Few people consider the internal- combustion engine environmentally friendly. Larry Mianzo could help change that. Mianzo is a key player in the auto industry’s efforts to build cleaner, more efficient engines. His innovations could usher in something called “electromagnetic variable valve timing.” In an automobile engine, valves are opened and closed in a fixed pattern by rotating camshafts. Eliminating the cams and moving each valve with an electromagnetic actuator allows optimal control of valve timing, ending power losses and providing a more tightly controlled combustion temperature. The result: a 15 to 25 percent boost in fuel economy and dramatically lower emissions. Mianzo has been working on automobile controls for the past eight years—first for Ford Motor, now for its spinoff Visteon, in Dearborn, MI. He has 12 patents issued or pending, including one on a novel road simulation strategy for testing new vehicles that has helped save Ford Motor $1 million a year. Mianzo has accomplished all this while earning a PhD and hasn’t missed a day of work since 1992.
Bill Nguyen is a serial entrepreneur. He led business or technical development in four startups before founding Onebox.com, a company that was among the first to provide e-mail, voice mail and fax access in a single mailbox over a conventional phone. Nguyen sold Onebox for $850 million in 2000,but not before hearing from wireless subscribers that they lacked similar data retrieval services. His solution? Seven Networks. Software from the Redwood City, CA, company lets customers of a wireless carrier access e-mail, voice mail and other Internet services simply by calling the carrier. Unlike subscribers to other wireless services, Seven customers don’t have to install extra hardware or software. So far, Cingular, Sprint PCS and Britain’s mmO2 have implemented Seven’s innovation, and Nguyen has secured $64 million in venture funding. His company has also partnered with Microsoft to create software that allows a company’s employees to wirelessly tap into its intranet. How does Nguyen do it all? For one thing, he is notorious for sleeping only three hours a day.
In Bangalore, India,six years ago, programmers working for overseas firms were commonplace. Innovative startups were not. That changed after Rajesh Reddy, trained at the Bangalore Institute of Technology, founded Gray Cell Technologies in 1996.Reddy ambushed a Motorola vice president on business in Bangalore and showed him that Gray Cell’s desktop- to-wireless network was more advanced than similar Motorola technology. Motorola soon licensed Gray Cell’s application for sending e-mail via cell phones and pagers and became Reddy’s first corporate customer. More U.S. investment in Indian information technology companies followed. By 1999 Reddy had renamed his company Unimobile and moved it to Silicon Valley, where it operated a wireless network connecting 370 carriers in 130 countries. The dot-com bust crippled the company, though. By summer 2001 Reddy was back in Bangalore launching July Systems, to create software that integrates wireless networks and devices into a global superstructure. With the backing of investor Ashok Narasimhan, and with business lessons learned, Reddy is confident July Systems will become a significant player this summer.
Rules “get in the way most of the time,” says cryptographer Vincent Rijmen. Last year the U.S.government chose the Belgian citizen’s encryption algorithm, Rijndael (pronounced rain-doll),as its new Advanced Encryption Standard. Rijndael replaced the aging, no longer unbreakable Data Encryption Standard, used since 1977 by U.S. government agencies and companies to safeguard everything from e-mail to phone calls. It beat submissions from many large competitors, including IBM, and will be widely used. Rijmen created Rijndael with Joan Daemen,36,a fellow postdoc at Katholieke Universiteit Leuven in Belgium. The duo pulled off the upset in part by throwing away what Rijmen calls a cryptography “rule”: to be secure, an encryption algorithm has to be exceedingly complex. Advanced computers would need trillions of years to decrypt information encrypted using Rijndael—yet the algorithm can run on devices like smart cards. Already, manufacturers plan to include Rijndael in cell phones, credit cards and Web browsers. “People will be using it without ever knowing,” says Rijmen, who recently became chief cryptographer at Cryptomathic, an Aarhus, Denmark, security firm.
Thanks to Jonathan Rosenberg, the Internet could usurp the role of the old-fashioned phone network. The key is a set of computer instructions that make it practical for the Internet to carry not just data but two-way telephone calls, teleconferences and pages. This “session initiation protocol” also supports new-fangled connections like instant messaging and “presence,” which tracks who is available online at any given moment. Rosenberg produced the protocol with Columbia University telecom expert Henning Schulzrinne while working toward his doctorate at Columbia and overseeing video compression research at Bell Labs. The telecom industry heralded the protocol, and the Third-Generation Partnership Project, a high-profile colloquium for setting wireless standards, adopted it in 2000.As chief scientist at East Hanover, NJ, startup dynamicsoft, Rosenberg has since been cooking up a suite of related software that would enable wireless phones to download voice, text and video and would let company Web sites provide voice links to live customer service representatives.
Inside an airliner, vibration frays a tiny piece of insulation, exposing an electrical wire; an arc of electricity ignites vaporized fuel—and a disaster. That’s what investigators suspect caused the 1996 explosion of TWA Flight 800. Electrical engineer Steven Shaw wants to make sure it doesn’t happen again. While pursuing his PhD, Shaw wrote algorithms that allow sensors to interpret minute fluctuations along every electrical line in an aircraft or building. This information can help building managers find faulty equipment or wiring and help airplane inspectors pinpoint electrical malfunctions—before problems turn deadly. Now a professor at Montana State University, Shaw is equally adept at theorizing, coding and working in the machine shop. The California Energy Commission is testing Shaw’s advanced load-monitoring systems on several state buildings. Better information about electrical flow can help building managers decide when to fire up backup batteries, fuel cells or expensive gas turbines.
“The most interesting thing in the world,” says University of Michigan professor and dirt biker Anna Stefanopoulou, “is balancing trade-offs to control complex systems.” Stefanopoulou works on electronic valves that could boost the fuel economy of conventional car engines by an estimated 10 percent and make practical exotic designs that are 30 percent more fuel efficient and free of nitrogen oxide emissions. A conventional engine regulates power with a throttle that controls airflow into cylinders; the timing of valves stays mechanically fixed. But the timing of electronic valves can vary infinitely, allowing the engine to “gain torque so fast it can break the crankshaft,” Stefanopoulou says. The native of Greece is developing such controls using sophisticated mathematical modeling, while high-end car companies “rely more on intuition,” says her Michigan colleague Jessy Grizzle. Stefanopoulou is also devising automated gears that would use engine compression to brake vehicles. She was already modeling the control of cars powered by fuel cells when the Bush administration dismantled an initiative to develop hybrid gasoline-electric vehicles–in favor of fuel cell power.
People often ask Lisa Su why she works for IBM—after all, aren’t startups where the glamour is? Su’s response: “I can run a group that’s like a startup, yet I have the resources available at IBM.” Her Emerging Products group focuses on low-power and broadband semiconductors as well as biochips. Its first product is a microprocessor that improves battery life in handheld assistants and cell phones. Su hired the group’s 10 employees and says their role is to develop broadband products that will “give my mom instant, unlimited access to information, anytime, anywhere, in any form.” After joining IBM in 1995,Su, who has a PhD in electrical engineering, played a critical role in integrating copper connections into semiconductor chips, solving the problem of preventing copper impurities from contaminating the devices during production. The technology, unveiled in 1998,led to chips that were 10 to 20 percent faster than those with conventional aluminum connections. Su showed she had management acumen and was allowed to start Emerging Products. “Lisa became an IBM executive in five years,” says colleague Scottie Ginn, “quicker than anyone I’ve ever seen.”
Iran native Vahid Tarokh works so quickly that by the time people apply his advances, he is often on to something else. Such is the case with his breakthrough codes to improve the speed, capacity and clarity of wireless voice and data communications. He developed the codes in 1996 at AT&T Labs, yet U.S. and international telecom standards bodies didn’t adopt them until 1999.Tarokh’s codes solved the problem of how best to get a signal from a base station to a cellular phone without fading. Solutions proposed by others, such as adding an extra antenna to the phone or sending the same signal on different frequencies, weren’t practical, so he created algorithms whereby multiple antennas at the base station could send the same signal simultaneously on the same frequency. For two months Tarokh worked day and night handcrafting his solution on huge sheets of paper. He moved to MIT in 2000 to work on “orthogonal frequency division multiplexing,” an advanced scheme for wide- band wireless communications. This summer Tarokh joins Harvard University as an electrical engineering professor.
In 1995,Steve Tuecke’s boss, Ian Foster, offered the organizers of a supercomputing convention a demonstration of “grid computing”— linking supercomputers at university and government labs into a single shared resource. The problem: the labs’ computers had incompatible hardware, security arrangements and queuing procedures, and nobody had written a program to resolve them. “My first thought was ‘Oh,jeez,’” says Tuecke, a software designer in Foster’s Distributed Systems Lab in Argonne, IL. But within weeks, Tuecke had created the code. The demo was the talk of the convention. Tuecke’s software grew into the Globus Toolkit—the “middleware” now used by hundreds of scientists worldwide to share high-end computers, databases and instruments remotely. Using Globus, a European Space Agency researcher could log into his desktop and run a climate simulation on a NASA supercomputer in California.Companies like Compaq, Fujitsu, IBM and Microsoft are eyeing Globus as a foundation for business- to-business Web services. “What the Web did for document sharing, the grid is doing for more general resource sharing,” notes Tuecke, who recently became the lab’s new software architect.
Media artist Camille Utterback’s award-winning video tracking exhibits create spaces where computers follow and interact with a person’s entire body. InText Rain, ademonstration based on patent-pending software created by Utterback and artist Romy Achituv, participants see themselves projected in real time on a wall while letters from the lines of a poem rain on their bodies. As the people move, the letters adjust accordingly. In Crossing, what appears to be an abstract painting on the wall is really a projection that ripples in response to a viewer’s movements. Utterback’s goal, both as an artist and an assistant professor at New York University and the Parsons School of Design, is to “help people realize that when technology systems are designed well, they are really fun.” Utterback, who in November 2000 started her own company, Creative Nerve, is a rare example of a computer programmer trained in the fine arts. Carl Goodman, curator of digital media at the American Museum of the Moving Image, says Utterback excels at following her curiosity and that her work “will stand up to scrutiny in the future, when the technology she’s using will no longer be novel.”
Les Welch has been fixing equipment ever since his bike broke in sixth grade. At aerospace giant Lockheed Martin he fixes manufacturing. Welch is applying lean-production techniques pioneered for auto assembly to the manufacture of F-22 fighter planes, bucking the defense industry’s history of inefficient production. Traditionally, aircraft have been built in one spot, with assembly workers walking many kilometers (this has been measured) to fetch thousands of tools and parts. In Welch’s approach, a nascent air- craft will move from one work center to another, each designed to minimize worker movement and maximize assembly convenience. Inventory is also reduced. Eric Ouellette at Lockheed Martin says the changes will cut manufacturing time up to 40 percent. Ouellette, formerly Welch’s senior manager, says Welch is “passionate about eliminating waste.” Before earning his industrial engineering degree, Welch ran manufacturing for a large family business that made aluminum toolboxes for pickup trucks. As for his personal tinkering, it’s now directed at his 1986 Jeep Grand Wagoneer.
Sean Willems’s quest for simplicity dates back to childhood; he recalls being “stressed that ‘Kansas’ and ‘Arkansas’ weren’t pronounced the same way.” Now the cofounder and chief scientist of Boston-based Optiant cuts through complexity as an industrial pioneer. Willems, who is also an assistant professor at Boston University, creates software that streamlines the flow of parts and materials to manufacturers. While a PhD candidate at MIT’s Sloan School of Management, he wrote an algorithm that optimizes such supply chain flows so manufacturers can cut costs by paring inventory. Previous algorithms solved only pieces of the problem; Willems addressed the task in its entirety. Willems tested his theories with leaders like Hewlett-Packard and Nortel Networks. When he applied them in a division of Eastman Kodak, the company cut inventory levels by 40 percent and saved $10 million over two years. The “cool thing about the life I lead,” he says, “is that I get to develop the theory at Boston University and apply it at Optiant.