Future Tense for Apr. 03, 2006
If I wasn’t sitting here in front of a computer, writing this column, I would be helping my wife with spring cleaning.
Honest, I would! I’m not fond of housekeeping chores, but someone has to do them and there’s no reason why my wife should have to work alone. Of course, I wouldn’t mind if technology could take on some of the load.
Why don’t we have more robots designed for simple household tasks?
The late Isaac Asimov, one of the grand masters of science fiction, described consumer-level robots more than 60 years ago in a series of short stories published in “Super Science Stories” and “Astounding Science Fiction” between 1940 and 1950 and compiled in the book “I, Robot” published in 1950, one year after I was born.
Asimov’s fiction inspired thousands of scientists and engineers who developed the technology of robotics (a word that Asimov coined in his 1941 story “Liar!” more than 20 years after playwright Karel Capek’s invented the word “robot”). No one has invented the “positronic brain” that guided Asimov’s robots and limited their behavior under his “Three Laws of Robotics” (first listed in 1942’s “Runaround”). However, the Defense Advanced Research Project Agency’s “Grand Challenge” has shown how computers can be linked together to manage complex tasks, somewhat like a human brain works.
In its “Grand Challenge,” DARPA offered a $1-million prize to the team whose unmanned, self-guided robot vehicle could complete a 150-mile off-road course through the Mojave Desert from near Barstow, Calif., to Nevada in 10 hours or less. More than 100 teams registered for the 2004 race. None came close to finishing. The best went about 7.4 miles.
Afterwards, some “experts” said it might take 15 years to solve all of the problems.
In fact, it took one year.
In the 2005 Grand Challenge, the Stanford (University) Racing Team’s “Stanley,” a Volkswagen Touareg SUV, won the race ó and its $2-million prize ó in less than seven hours. Four other vehicles completed the 132-mile course. All but one of the 23 competitors went beyond the 2004 mark.
Stanley’s “brain” was six Pentium computers. For “sight,” it had five laser range-finders, two radars and a digital vision system. Stanley also had a GPS receiver, a GPS compass, an inertial guidance system and a precise digital odometer.
No one really knows how much it cost to build, because corporate sponsors donated much of the hardware. Most people on the 60-member team were either graduate students who work cheap or undergrads who cost nothing at all. Still, they beat the Grand Challenge with only 15 months’ work. A government contractor probably would have taken 10 years and $150 million to do the same job.
However, I’m not looking for a car that can drive itself from Cana to Charlotte. All I need is a robot that can wash windows, clean the gutters, air out the linens and vacuum the floors.
As it happens, the last exists. Even better, it’s affordable.
Two MIT computer-science students, Helen Greiner and Colin Angle, and their professor, Rod Brooks, started the iRobot company to develop low-cost robots for common consumer uses. In September 2002, they introduced Roomba, a self-propelled, self-navigating vacuum cleaner. Since then, they’ve sold about 1.5 million copies of the “robotic floorvac,” now in its third generation, and recently introduced Scooba, a self-guided floor mop.
A basic iRobot Roomba costs $150; a top-of-the-line Roomba Scheduler is $330; there are three intermediate models priced at $200 to $300. All are about 13 inches in diameter and less than 4 inches high.
Roomba operates on rechargeable batteries. They do not require special programming. They could hardly be simpler. The owner puts Roomba in a room, pushes the “on” button and selects “clean,” “spot” or “max.” The Scheduler model can be preset to start at a later time ó for example, while the owner is at work or asleep.
When Roomba starts doing its thing, the device doesn’t “see” a room or its contents. Instead, Roomba’s computer applies certain strategies, defined in software, such as moving at random until it finds a wall, from which point it circles the room and then spirals in toward the center. If it does not find a wall in a certain period of time, the little robot uses a random pattern to cover the floor. Roomba automatically tracks how much dirt it collects and learns to concentrate on certain areas. If power runs low, it stops and chimes an alarm or, in newer models, follows an infrared signal back to its recharging unit. When it finishes with a room, Roomba signals its owner to empty the dustbin.
Roomba doesn’t work well on shag or thick-pile carpets and its wheels can be clogged by dog or cat hair (a problem that goes away after a few uses). Roomba goes under most furniture, including beds, but occasionally gets pinned down and calls out a distress call (newer models come with a Virtual Wall device that tells Roomba to stay out of certain places). It’s hard-working, but dumb, and takes longer to vacuum a room than I would, but I’m not doing the labor, so what do I care?
Roomba has few competitors. Elextrolux, the vacuum-cleaner company, demonstrated a similar system in 1997, but only sold the Trilobite in Europe until it recently introduced an $1,800 model in the U.S. market. The Trilobite reportedly has more computing power, but refuses to go near walls. I’d like to give you more details, but the company’s Web site wouldn’t come up (never a good sign).
Helen Greiner, who’s now chairman of iRobot, believes Roomba has been successful because it is inexpensive compared to its value to an owner.
At $150, Roomba certainly seems reasonably priced.
Do you suppose my wife would prefer an iRobot Roomba or a ring for our anniversary ... in the future?