If you ask a robot scientist why his industry is so interested in cleaning our floors, chances are he'll mention the "three D's." Robots, it turns out, are best suited to replacing humans in jobs that are dirty, dull, or dangerous. Sweeping, vacuuming, and mopping fit the first two D's, and depending on your dog's diet, perhaps the last one as well. It's no wonder, then, that the world had such high hopes for the Roomba, the autonomous floor-cleaning robot that first went on sale in 2002. The Roomba was built by iRobot, a company best known for its bomb-disposal machines —in other words, excellent preparation for designing a tool to help frat brothers clean up after keggers. The Roomba quickly became an icon— nerds hacked it, cats loved it, and lonely people grew to consider the bot something like a pet. There was just one problem: When it came to cleaning, the Roomba sucked, and not always in a good way.
The Roomba's creator, iRobot, has always been upfront about this. The company bills the machine as a complement to manual floor maintenance, not a replacement. Getting a Roomba is not going to eliminate the need to sweep or vacuum your house—it's just going to let you do so less often or assuage some of your guilt at never doing so at all. In my experience, though, the Roomba has struggled to meet even this low bar. When I first got the robot in 2007, I loved it—if I used it per the company's directions, it cleaned up quite well. But over time, I grew to resent the machine. The Roomba, I found, was neither very bright nor obedient. If I didn't clean up the room before I turned it on—moving chairs, setting up "virtual walls" to restrict its movements, removing errant video cables and other wires—the Roomba would either get stuck or wander around aimlessly. I've taken to using the Roomba less and less—it's both quicker and more effective to get out a broom or Dustbuster.
Perhaps because of these flaws, iRobot's recent Roomba sales haven't been great. Home robot revenues fell in 2009, and domestic sales have been flat over the last few years, with most growth coming from an international expansion. At least in America, it seems we've soured on the Roomba.
Despite the Roomba's problems, competing robot cleaners haven't done very well; the Roomba remains the best-selling robotic floor cleaner in the world. In the last few weeks, though, I've been testing two new robots that could well threaten its preeminence. One of them is a vacuum cleaner called the Neato, which sells for $399. (The Roomba, by comparison, runs from $199 to $599, depending on the model.) The other, the Evolution Mint, is a $249 sweeper and mop—that is, it works on tile, wood, and other hard floors, not on carpet. I found both the Neato and the Mint to be pretty good cleaners. In my tests, they seemed to get floors cleaner—and required less regular maintenance and pre-cleaning—than the Roomba. Still, neither one fulfilled the dream of those of us who loathe housework—they couldn't clean well enough to eliminate the need for human help. Like the Roomba, they're both supplements to a broom, a mop, and a vacuum, not a replacement.
To understand what makes these new gadgets superior to the first generation of robo-cleaners, it helps to understand how the Roomba works. The iRobot company was co-founded by Rodney Brooks, an MIT roboticist who has long been obsessed by the movements of insects. (He's profiled in the brilliant Errol Morris documentary Fast, Cheap & Out of Control.) In the 1980s, Brooks provided a key insight to the emerging field of artificial intelligence—instead of trying to create machines that think the same way you and I do, he argued, we'd have more success creating machines that operate according to simple reflexes, like insects. That's the Roomba's operating philosophy. As the book Hacking Roomba explains: "Roomba has no room map or route plan. It has no overall view of what it is doing. Instead it functions much more like an insect: going toward things it likes (dirt, power) and away from things it dislikes (walls, stairs), moving in predefined movement routines while occasionally and randomly jumping out of a predefined routine."
The problem with this approach, Roomba's competitors say, is that it's wasteful. The Roomba cleans a floor the way you would if you were blindfolded—it moves in one direction until it hits something, then turns in another direction and moves until it hits something else. Eventually it cleans every part of the room—but the anti-Roomba crowd claims that it does so unevenly, going over some parts of a room many times while cleaning other spots just once. A more systematic approach could yield greater efficiency: If a robot cleaned each part of the floor just once, it would have a lot more battery power to clean more forcefully and could clean more quickly, to boot.
That's the theory behind both the Neato and the Mint. "We clean your floor the way a Zamboni would, or the way you would," says Max Safai, the CEO of Neato Robotics. The Neato uses several different sensors to create an internal map of a room. Based on this map, it will first clean the room's perimeter before going back and forth within the perimeter in a systematic way. To see what I mean, watch this clip of the Neato tooling around my dining room:
Because this back-and-forth pattern is more efficient than Roomba's blind walk, the Neato can afford to pack a much more powerful vacuum cleaner onboard. As a result, the Neato left a discernible, just-cleaned pattern over my thick-pile rug—the sort of pattern I get when I use an upright vacuum, but that the Roomba could never achieve. What's more, because the Neato creates a map as it goes, it always knows where it is. This allows it to go from room to room, cleaning each area before it moves on to the next one. When it runs low on power during its cleaning mission, it can pilot back to its charging base and then start back up again from where it left off. Safai says that after forthcoming software updates (the Neato has a USB port that allows you to install new firmware), it will be able to perform a couple of more tricks—remembering and avoiding places where it's had trouble navigating and allowing you to schedule cleaning certain rooms on certain days.