Unlike nature, the media adores a vacuum.

Specifically, the media adores Colin Angle's vacuum, Roomba, a sleek silver saucer that last autumn whirred its way onto the year's best-products lists of Time, Business Week, and USA Today, among others. With the publicity helping to drive sales of tens of thousands of units over the holidays, Roomba, from iRobot Inc., had the kind of product launch most entrepreneurs would sell their grandmother up the river for. "We were No. 6 on Entertainment Weekly's ranking of hot topics, right under Angelina Jolie and Prince Harry," says CEO Angle, awe vying with amusement in his voice. "Now that is powerful."

It's also an awful lot of heavy breathing for a category that traditionally has delivered all the sex appeal of baggy support hose. Roomba's va-voom quotient derives from the fact that it's a robot: It uses artificial intelligence and infrared sensors to sweep rooms on its own, without requiring some weary soul to slog along behind it. The potential market includes seniors, the disabled, and anyone who lives someplace that has floors. "When the fertility drugs kicked in and my wife gave birth to quadruplets I knew we'd have our hands full keeping the house clean," confides one reviewer on Amazon.com. "This little marvel tools around the house providing an unceasing blanket-sweep of our floors."

The vacuum is helping its makers clean up as well. Roomba is the first offering from the nascent consumer products division of iRobot, which expects to top $50 million in sales this year, compared with $15 million in 2002. The company, based in Burlington, Mass., has also raised $27.5 million in five rounds of venture funding, most of it after the technology market fizzled. "They've got sales, they've got momentum, which in this economy is staggering," says one private equity investor who is not involved with the company but is familiar with its plans. "They've had a huge hit with Roomba, and that attracts a lot of people."

Until recently, however, iRobot looked nothing like a mass-marketing sensation waiting to happen. For 13 years it built its reputation as a boutique engineering contractor, designing products that were too cool for schools of them ever to be produced. There was part of a teeny-tiny robot -- less than a centimeter long -- developed for the Japanese Ministry of Trade. A crab-walking minesweeper for the Department of Defense. A rugged oil-well-repair bot for Baker Hughes (now Halliburton). The company was an idyllic environment for engineers eager to grow and learn and not get rich. "We were getting contracts from people who wanted us to do something they found interesting but only wanted one of," says Rodney Brooks, iRobot's chairman and cofounder. "It was like being an artist working on commission."

But iRobot's founders had this dream: to place in every home robots that are cheap and task-specific and labor together with the unfussy coordination of Snow White's dwarves to clean your carpets, and perhaps in the future, wash your windows, mow your lawn, or mop your floors. It sounded like a job for a consumer-products giant like Sony, not a 100-person firm with no manufacturing experience, marketing exper-tise, channel contacts, or understanding of consumers. Perhaps the most daunting obstacle was iRobot's techno-elitist culture. "When Hasbro asked us to make a robotic velociraptor to tie in with Jurassic Park III, the engineers came to me and said, 'We can't possibly work on that. It's too dull,'" recalls Angle.

If the shift to mass marketing is not done well, it can mean death to the company that bungles it.

Many innovative tech companies have stumbled lately on the way from conception to market. Others, like digital-video-recorder pioneer Tivo, struggle to transform their ideas into a marketing message everyone "gets." Meanwhile, small R&D shops that supply brand-name manufacturers rarely break out on their own -- and for a reason. The shift iRobot had to make, from high-cost prototype design to every-penny-counts mass production, "is a hugely daunting transition," says Robert Bruner, executive director of the Batten Institute at University of Virginia's Darden Graduate School of Business. "If it's not done well, it can lead to setbacks or even the demise of the company that bungles it."

Angle, 37, embodies the argument for nurture over nature.

The boyish, lank-haired CEO was raised by his mother and engineer stepfather: three stepbrothers also became engineers. His biological parents, brother, and two half brothers, by contrast, don't count a gearhead among them. Growing up in Schenectady, N.Y., Angle displayed early signs of what he would become, building pinball machines and crossbows and erecting elaborate pulley systems in the trees of his backyard. "When I was two years old you could put me near a stream and I'd dam it," says Angle. "I invented thousands of things, none of them useful."

At MIT in the late 1980s, Angle quickly succumbed to the gravitational pull of Rodney Brooks, director of the university's artificial intelligence lab. In the field of robotics, Brooks was famously controversial until he turned out to be famously right about many fundamentals of the science. Most notably, he argued that robots need not think like humans to be useful, and he and his students went about proving it by building an assortment of insectlike machines -- imbued with reflexes rather than reasoning -- that could move around and perform simple tasks.

Like Roomba more than a decade later, the insect robots piqued the media's fancy. Angle's senior project, an ambulatory critter called Genghis, made the cover of Popular Science before retiring to the Smithsonian. Brooks, meanwhile, was featured in Errol Morris's documentary Fast, Cheap & Out of Control, a panegyric to obsession and human striving in which he shared screen time with a lion tamer, a topiary gardener, and a connoisseur of naked mole rats.

In 1990, Brooks -- then 35 -- and Angle used credit cards and bank debt to launch iRobot from a small apartment in Somerville, Mass. (Helen Greiner, another graduate of the AI Lab, joined them shortly thereafter.) Brooks assumed the title of chairman, and Angle declared himself president. Forgoing sleep and salaries and putting their personal lives on hold, the founders pursued one overarching goal: to build robots that would touch people's lives. But at first the only lives touched were those of other students of robotics, who used the start-up's products for their own experiments.

Now, as growth markets go, building robots for robotics-studies programs isn't one. Soon iRobot was seeking government contracts and, later, corporate work. A list of its projects over the next decade is as eclectic and lacking in direction as a freshman's course schedule. The company's growing cadre of engineers built nuclear-waste detectors and industrial floor waxers and timid toy balls that whimpered when rolled into dark corners. Each project became a division; those divisions multiplied to as many as 12, dwindled to a handful, and then swelled as customers chose to manufacture, scrap, or back-burner their prototypes. The goal of such diversity was to mitigate risk. "We had our fingers in all these pies because we had no killer pie," Angle says. "And we didn't know how to decide what a killer pie was other than to let Darwin do it.

"Focusing this company too early," says the CEO, "would have killed it."

The eclecticism of iRobot may have saved it from the fate of some Internet companies, which assumed the mere existence of their technology would somehow will a market into being. But it also allowed the business to store up a trove of intellectual property. The importance of IP was brought home early to Angle, who in 1992 negotiated away the rights to a robot called Grendel to one of his customers. "We had to segregate that whole product line and make sure that other derivative products followed a different technology path so that [the customer] wouldn't end up owning a big chunk of what we did," says Angle. "It was very scary stuff, but it was a good early lesson."

It's a lesson many never learn: Small contractors routinely cede valuable patents and trade secrets to their larger customers. After Grendel, however, iRobot insisted on retaining the rights to all technology developed during a project, except for a few well-defined applications specified by the client. Most customers were willing to go along with that, so long as they could benefit from R&D done on someone else's dime. "The ability to draw on IP developed elsewhere became part of the story we tell, part of what we sell," says Angle. It also left the company free to use the fruits of its earlier labors -- specifically, cleaning technology developed for Johnson Wax Professional and tiny processors created for Hasbro -- in Roomba and its anticipated sister products.

The company's diversity made it flexible; its intellectual capital made it strong. But to bring its robots to a wider audience, iRobot needed to be one more thing: cheap. For a contractor used to spending tens -- sometimes hundreds -- of thousands of dollars on envelope-shoving prototypes, that wasn't easy. "This is my monument to not understanding the realities of the world," says Angle, taking down from a shelf a plastic-molded tableau in which a half dozen cartoonishly costumed figures surround what looks like a child's book. "It's a storytelling machine. We developed a scripting language so that you could write something -- 'King Croft turned to Jenny and said, "I'll get you"' -- and the characters would actually turn and make the right gestures and the lights would come up and there would be dialogue. It was unbelievably cool."

The management at Hasbro -- to whom Angle presented the prototype in 1996 -- agreed. But then Angle opened it up and showed off the innards. "The unit price on the microprocessors was $60 each," says the CEO. "This flash card was something like $400. Just the parts were $3,000. We thought there was some magical thing where you could give this to the toy industry, and they would say, 'Here's your royalty,' and it would turn up on the shelves at Toys "R" Us for $19.95.

"We were profoundly shot down," he adds unnecessarily.

Still, in 1998 Hasbro agreed to fund an R&D team at iRobot and became its exclusive partner in the toy industry. For two years, as his engineers pitched ideas for toy concepts and technologies, Angle got religion on cost control. The catalyst was rejection: dozens of thumbs-downs on technically elegant -- but way too expensive -- projects such as board games that interacted with players and balls and bats that understood how they were being used. "The first time it happened it was unbearably painful," says Angle. "By the fifth time, the engineers learned to maintain some emotional distance, which allowed them to broaden their attention as to what makes the project rewarding."

What makes the project rewarding, the toy group decided, was seeing it on a store shelf. One product -- the emotionally responsive doll My Real Baby -- finally made it to market in 2000; iRobot technologies turned up in a variety of the toymaker's offerings as well. Those all came about because the engineers promoted cost to priority one. They sourced materials aggressively. They learned to make robotics with plastics rather than with expensive machined aluminum. Detailed bills of material began showing up in every product pitch. To keep itself focused, the group adopted a mantra: "We have no money."

As iRobot pitched ideas for toy concepts, Angle got religion on cost control. The catalyst was rejection.

Today, Angle waxes evangelical on the subject. "When people say to me, 'I'm thinking of making a consumer product,' I say, 'Have you had a knock-down drag-out fight over a nickel yet?'" says the CEO. "They either give me a mystified look or they say, 'No, Colin, it's actually a penny.' And if they say it's a penny, I know they understand the consumer-products business."

It was Angle's own grasp of the consumer-products business that led, in 2000, to the idea for Roomba. At first the founders assumed they'd find a partner, like Hoover, to manufacture and sell the device. But looking back at all they had learned from their customers -- how to source materials and ensure safety and manufacture cheaply and design for the mass market and work with companies in the Far East -- "we decided to do it ourselves," says Angle. "Because we could."

Barbara Bund, a senior lecturer at the MIT Sloan School of Management, tells the story of a group of engineers at Quaker State who were invited to observe consumer focus groups discussing the company's motor oil. "This one engineer was sitting in the backroom watching a group in which not one single person understood the difference between synthetic oil and nonsynthetic oil," Bund says. "He walked out of there and said, 'We've got to get smarter focus groups.'"

It's an attitude Angle knows well. As the CEO began carving out the consumer division in 2001, he considered that his crew of tech huggers might run into problems. The company's engineers (60 out of a staff of 100) are a preternaturally brainy bunch: A third hail from MIT, where they worked on everything from ocean engineering to carbon fiber solar cars. Although such people have their own Uncle Morts and Auntie Mabels, they do not ordinarily build stuff for them. But the vacuum-buying public, Angle knew, wouldn't tolerate a learning curve steeper than a flat line.

To prevent a mass case of the cultural bends, Angle seeded the consumer robotics group with five engineers from the toy division, and then had the team observe focus groups of ordinary people who clean things. As engineers worked their way through 20 versions of the product, they brought each iteration home for testing by spouses, friends, and neighbors. The marketing staff also auditioned Roomba in their houses and filled out surveys reflecting the average-joe perspective.

The iRobot engineers responded to the feedback with enthusiasm. "We used to talk about training users to operate the robot, and now that's something that we never want to do," says Jim Lynch, one of the chief designers of Roomba. "We have a picture of this big button that says clean. There were many times someone would grab that piece of paper and put it on the table and say, 'That's the user interface we want. That's how simple we want this thing to be.'"

The focus groups also reinforced the lesson that consumers don't like products that tax their wallets any more than ones that tax their brains. Roomba's $200 price tag is a key defense against competitors. Because only a few of Roomba's manager-engineers learned cost control at Hasbro's knee, the company's new chief financial officer, Geoffrey Clear, spends hours helping them anatomize bills of materials and pointing out where, say, an 8¢ resistor got preferential treatment over a 3¢ alternative.

Clear's hiring -- he's the company's first CFO with a portfolio broader than government contracts -- was another marker along iRobot's road to mass-market-dom. Formerly an executive at Micro-Touch and 3M, he's helping iRobot forecast demand, plan production, and perform all the other financial chores that contract engineers don't worry their pretty little heads about. The CFO also has experience preparing companies to go public -- an outcome becoming legible in iRobot's tea leaves.

In manufacturing, too, the emphasis has been on building iRobot's mass-market chops: The company hired a former Hasbro employee to help establish its Far East operations and a veteran of action-figure producer Toy Biz to run them. And in March, executive vice president Greg White, who hails from appliance maker the Holmes Group, came on board to assume the consumer-products reins and develop iRobot's brand.

Not that iRobot is smarting from its earlier expertise deficit. In fact, the company's first sallies into distribution and marketing were shrewdly executed. It launched Roomba in September 2002 -- in time for the Christmas season -- through Sharper Image, Hammacher Schlemmer, and Brookstone, relying on their appeal to gift buyers and willingness to demonstrate products. In February Roomba expanded into Bed Bath & Beyond and Linens 'n Things and, in April, Target. Those venues appealed, Angle says, because of their willingness to run tapes of the product in action, offer demonstrations, and provide other support more typical of specialty shops.

Meanwhile, a publicity blitz helped build consumer demand. Media Strategies, a Boston PR firm, pitched Roomba's story aggressively to one outlet in each print category: a newsweekly (Time), a national daily (The Wall Street Journal), a technology book (PC Magazine), and so on, sending demo versions to a few chosen reporters. On launch day Angle did 22 TV interviews via satellite from his office. Once Time and shows like Live With Regis and Kelly established that Roomba was news, iRobot swept onto the women's and bridal magazines, the favored distractions of its ultimate audience. "We even got the Good Housekeeping Seal of Approval," says Angle. "How retro is that?"

Yet, the deluge of Jetsonian headlines celebrating Roomba's arrival presented iRobot with an unusual challenge. The company needs Roomba to be first and foremost an appliance: an attractive alternative to uprights and dust-busters, practical to the point of mundaneness. To be seen as innovative is grand: to be a novelty is dangerous. The last thing Angle wants is for Roomba to become the Internet refrigerator of vacuum cleaners.

For that reason, the product's robotic pedigree was potentially troubling. It's no accident that the word robot -- evocative of homicidal cyborgs and space-age valets -- isn't uttered once in the company's 30-minute infomercial. Nor does it appear on the first iteration of the product's packaging, except as part of iRobot's name. Instead, the company describes its baby as an "intelligent floorvac system." "Most people would go to the market saying, 'We're the best in the world at artificial intelligence! We know more than you do and robots are the answer!'" says MIT's Bund. "Instead they're using consumers' terms. They're not letting their technical egos get in the way of communicating."

But obfuscation is already on the wane. Reassured by letters, e-mail, and Web forum comments from customers largely tickled by the product's cyber self (60% of Roomba owners name them, says Angle), the company will use the word robot in packaging hitting the shelves this summer. And in the months to come it will ramp up marketing campaigns aimed at explaining and, ultimately, celebrating its robot technology. The reason for the shift is impending competition. Electrolux, for one, has developed a robot vacuum that has so far been kept at bay by its high price (around $1,500). Once competitive products get cheaper, iRobot can win only if consumers have learned to value robotic characteristics and, thus, prefer to buy vacuums from robot experts rather than robots from vacuum experts. "As competitors come in, our amount of robot expertise equity becomes more important," says White.

Even as it has funneled resources Roomba-ward, iRobot has not abandoned its contract industrial and military divisions, which Greiner runs. There, hard-core innovators can still be found building everything from self-driving vehicles to the burly Packbot reconnaissance robots that got their close-up in Afghanistan and Iraq. And, of course, all three divisions share technology among themselves at will.

That goes for blue-sky research as well. The federal Defense Advanced Research Projects Agency (DARPA), for example, is funding iRobot's Swarm project, designed to get robots to coordinate among themselves and act as a group. One recent morning senior lead research scientist James McLurkin and software engineer Jennifer Smith demonstrated the technology on a herd of softball-size robots resembling silver Dumpsters frosted with Christmas lights. At the command of McLurkin's controller, the lead robot, marked by an antenna, scuttled forward; it then sent a message commanding six more bots to fall into line behind it. "DARPA wants us to work on exploration behaviors," says McLurkin, "getting 100 of them to spread out, look for open space, stake out the ground, and find some object of interest." Applications include clearing mines and locating the wounded.

But they also include housework. Angle proposes using Swarm's programming in a line of cheap domestic robots. Such appliances could decide among themselves, for example, which gets first crack at the family room: the carpet sweeper or the waxer. And, of course, all of iRobot's appliances would be compatible with one another, a strategy that could potentially give iRobot a Microsoftian advantage. "Having the robots talk to each other allows us to take on more of the home-cleaning task $200 at a time," Angle says.

Angle, in other words, still envisions his robots touching the world -- the dusty, linty, crumb-crusted bits of it anyway. And now he is looking for fulfillment elsewhere too. With their management team in place and a steady stream of revenue, all three of iRobot's founders are, after more than a decade, reclaiming their personal lives. Greiner recently bought a house. Angle, who sacrificed his first marriage to the start-up, has an eight-month-old daughter with his second wife, a psychiatrist.

As for Brooks, he spends the equivalent of just one day a week at the company, devoting the rest of his time to the AI Lab, where he's moved onto projects more humanoid than entomological. Now that, after all these years, he's finally building robots that affect people's lives, does he still have a dream? "I want to build a robot that someone would feel morally aghast at switching off because it would be like killing a life form," says Brooks. "That's not a commercial dream. But it's my dream."

Leigh Buchanan (lbuchanan@inc.com) is a senior editor at Inc.