A burgeoning market in robotics is upon us, offering unprecedented opportunity for eager and willing innovators and entrepreneurs
See Dexter walk.
See Dexter jump.
See Dexter help usher in a brave new robotic world.
Inside the Mountain View, Calif., lab at Anybots Inc., and at dozens of labs like it around the world, robots are learning how to mimic what it took humans eons to perfect - walking, grasping and, more important, learning from mistakes.
Dexter is a bipedal droid standing 5 feet, 10 inches. Its joints, driven by air cylinders, are springy and flexible like human muscle. Dexter's legs constantly adjust based on its computerized sense of balance.
"The robot equivalent to your inner ear," says Anybots founder Trevor Blackwell. "It walks and balances the same way humans do."
Dexter also is learning to run.
Then there's Monty, Dexter's mechanical cousin. Instead of legs, Monty is set on a Segway-like chassis. Monty's mission is to master picking up stuff. One hand consists of five articulated digits; the other is a blue metal claw. In one demonstration, Monty makes coffee with a French press. It's a time-consuming, messy affair, with grounds and water spilling on the counter. But like Dexter's walking, Monty's attempts are baby steps to the future.
Although still far from the abilities of NS-5, the super android from the 2004 movie I, Robot, Dexter's and Monty's achievements represent the ongoing renaissance in robotics. From Genoa to Tokyo to the halls of Stanford and Berkeley, teams of roboticists are toying, tinkering and testing ways to make robots more human.
Consider Honda Motor Co.'s ASIMO, a well-traveled servant robot that regularly gets media exposure. The stormtrooper-white droid can walk and serve drinks on a tray. ASIMO (short for Advanced Step in Innovative Mobility) uses a walking algorithm called zero moment point, or ZMP. It's a geometrical, programmed constraint that guarantees stability.
Dexter and other breeds of dynamic-balance bots don't use ZMP. It had to "learn" to walk. And, as Blackwell explains, its learning software will soon lead to a much wider range of walking abilities than could ever be programmed.
True, natural walking, then, would allow robots to move more freely about our uneven, random human environment, where dog toys, steps and other "hazards" abound. Grasping would allow robots to manipulate that environment. And having the ability to learn through trial and error could put robots literally at the doorstep of mainstream consumers. Think what walking upright, the opposable thumb and a bigger brain did for our ancestors, and you get a glimpse of what's possible. Robots as maids, manual laborers and mechanical beings that will do all our mundane and dangerous bidding.
Metal slaves, to be crude about it.
"Inevitably, people in entry-level jobs will be displaced by the new machines," write (Gregory Benford and Elisabeth Malartre) in their book, Beyond Human: Living with Robots and Cyborgs. "Pool cleaners, golf caddies, and high schoolers hoping to cut lawns or shovel snow will not be needed; demand for unskilled illegal immigrants will ebb."
But this is no time to be squeamish. The robot market is poised to explode, as advances in hardware, software and even the understanding of our own physiology and cognitive processes expand.
ABI Research, an Oyster Bay, N.Y., technology consultant and research firm, forecasts that the personal or consumer robot market alone will be worth $15 billion by 2015. That's twice as big as the current computer-game market.
"Over the long term, expectations of what robots should be able to do will rise," ABI Research principal analyst Philip Solis says in a report. "The increasing cost of more complex servos and sensors will be balanced by the increasing value consumers place on robots, and a growing willingness by consumers to spend more on robots that can make their lives easier or more fun."
This also means opportunity for professional and independent innovators. Robotics synthesizes the worlds of software and hardware development and all manner of industries, including music.
Last year, legendary guitar-maker Gibson introduced the world's first robotic guitar that tunes itself. Price: $2,499. Analysts say the self-tuning Les Paul could make guitars more appealing at a time when musical instrument sales are falling and computer games are booming.
Industrial robots got their start in 1961, when General Motors installed the Unimate to weld auto bodies. Now, robots are ready to march further into other fields, from deep space to the deep sea.
Researchers with Norwegian oil company StatoilHydro are developing open-ocean petroleum platforms designed to be run by robots.
"An automated platform doesn't need personnel, and therefore neither does it need fire systems, sound insulation, catering or a whole range of other installations," Anders Røyrøy, project manager for research and development projects at StatoilHydro, told ScienceDaily. Human operators would control the automated platform from the safety of shore.
Futurists also envision autonomous harvesters to gather food, robotic soldiers to fight our wars and even (gasp) sexbots to satisfy certain carnal desires.
We're already familiar with single-function task robots. iRobot's squat, plate-sized Roomba vacuum cleaner perhaps is the most widely recognized. Introduced in 2001, it's undergone a variety of improvements, including an industrial model that can suck up wood chips, nuts and bolts. The company recently released the Looj, a robot that cleans gutters.
Robotic lawn mowers, or mowbots, hit the market in 2000. Today's high-end versions run longer and return to their docking stations when done mowing the yard. They cost between $1,400 and $2,500. Safety features also have improved. Friendly Robotics' RL 1000 Robomower stays within an area that's defined by a hidden perimeter wire. It also shuts off if it's lifted or turns over. Still, that hasn't stopped some contemplating the worst. One Internet wag said, "You'd think we would learn from the countless depictions of robots in the future, ROBOT + BLADES = BAD! Can't wait to hear about a story of this taking off someone's leg or something."
But the bigger potential arrives when robots become more integrated into the human social landscape.
For instance, the Stanford Artificial Intelligence Robot, or STAIR, project seeks to engineer a robot to perform a variety of tasks, according to Andrew Ng, a computer science professor leading the STAIR initiative.
Ask a 5-year-old to fetch you a green notebook off a table containing various notebooks of different colors, and chances are he'll retrieve the right one. But when it comes to robots, this otherwise simple command requires a dizzying array of computations, including color recognition, navigation and hand dexterity. To get to this level of sophistication, Ng says robots need "broad competence intelligence." Once there, he adds, robots with this basic intelligence could help the disabled or elderly around the house.
Japanese and South Korean researchers, in particular, are exploring the realm of robots as butlers and caregivers. Honda's ASIMO, which can recognize individual faces, is a prototype robo-servant. (It also has served as a demonstration and testing platform for automotive technology. ASIMO systems that help prevent it from tipping over have been adapted to help prevent vehicles from swerving. It also has been responsible for technology that warns drivers about impending collisions.)
Toyota, the world's largest car company, announced in December that service robots would soon become one of its core businesses. The government heavily subsidizes development of these machines.
Mitsubishi developed Wakamaru, a robot that the company hopes will form "meaningful relationships with human beings, initiating conversations with family members and offering services such as alarm, news, weather and e-mail dictation," according to the authors of Beyond Human.
Wakamaru can look after the house, provide video streams over cellular networks, and cull useful info off the Web while maintaining its own autonomous "rhythm of life."
And then there's Nuvo, a general-use robot made in Japan. It provides "empathy" for the aging in nursing homes.
"I don't like the idea of robots replacing people," says Blackwell, the founder of Anybots, and "father" of Dexter the walking bot and Monty the grabbing one. "I think robots should do boring, dangerous, dehumanizing jobs and let people interact with people."
That said, the 38-year-old doctorate and millionaire (he made a bundle selling his Internet storefront system Viaweb to Yahoo! in 1998) sounds like those guys who pine for flying cars.
"I thought by the time I was 30, there would be robots walking around everywhere," he says. Unsatisfied at the pace of robot advancements, he started Anybots in 2001.
Blackwell believes when it comes to robots, we're at just the right point of software and hardware development - that magic time just before a revolution, when technology and the market begin to mature. Kind of like the personal-computer era of the late 1970s.
If he's right, Blackwell's vision of robots doing our dirty work will arrive within a generation, if not sooner.
"The potential market for these things is on the order of computers," he says. "Whoever gets it right will be the next Microsoft. We're hoping it will be us."
