The Race For the Personal and Service Robot Market

Meet Wakamaru and Roomba, two householdhelper robots with very different pedigrees. Wakamaru, from Mitsubishi Heavy Industries, is a waist-high bot with a canary yellow exterior and limpid eyes. It can recognize 10,000 Japanese words, identify eight family members by face or voice, remind you to make an appointment or take your meds and, if somebody breaks into your house, send photographs of the intruder to your mobile phone. When the machine rolled off the assembly line in 2005, Mitsubishi expected U.S. sales to reach 10,000 models a year, despite the bot's $15,000 price tag. Instead, the company filled only a few dozen orders. Wakamaru is now off the market and being rented out as a receptionist at $1,000 a day.

Roomba, by contrast, looks more like an appliance than a robotic friend. The Frisbee-like disc's sole purpose is to vacuum, which it does automatically, thanks to sensors that adjust the settings to suit different floor types, avoid drop-offs like stairs and navigate between table legs and household pets. Starting price: $130. Massachusetts firm iRobot Corp. has sold more than 3 million of the machines.

Wakamaru and Roomba represent radically different approaches to the next big thing in robotics: the use of robot assistants in the office, hospital and home. The Japanese, who have long been fascinated by the robot as android, are concentrating on making machines that look and act like human beings. U.S. firms, on the other hand, have eschewed the flashier android approach and instead are emphasizing products that, like Roomba, are narrowly targeted to specific tasks like mowing lawns, cleaning pools and taking patients' vital signs.

So far, the success of Roomba suggests that the U.S. firms have the upper hand. But the race is only beginning and the stakes are potentially huge. The market for personal and service robots is about $3 billion now but is expected to reach $15 billion by 2015, according to the Japan Robotics Association and market analysts like ABI Research. In 10 years or so, experts predict, sales of personal robots could surpass sales of industrial robots, now about $4.6 billion a year.

The issue for robot developers is whether the technology of artificial intelligence will allow Japanese developers to fulfill their vision of friendly robots capable of working alongside people. If so, Japan could be in a position to dominate the next phase of robotics. If not, the Americans, with their pragmatic but uninspiring designs, could win the race.

Japan approaches this new market from a position of strength. Over the past 50 years, it has become the undisputed leader in industrial robots, supplying 40 percent of the world market. At the same time, Japanese pop culture has become saturated with images of friendly droids from manga cartoons and animé, and bots by Sony and Honda are as famous in Tokyo as Jessica Simpson is in Texas. Japan's robot industry—with the help of $100 million in research funding from the government—is driven in large part by the dream of a day when droids will aid humans in almost every aspect of daily life.

There's the egg-shaped PaPeRo—recently rated the most popular bot in Japan by Robot Life magazine—which works select day-care centers, singing songs and reading e-mails to children according to texted instructions from parents. There's Actroid, a mannequinesque gynoid who wows corporate guests with her dynamic facial expressions and cheeky conversation skills (ask her how much she weighs, and she'll tell you what she can bench-press).

The Tmsuk Co. makes a cheery yellow and white bot that watches shoppers' kids at a mall in Fukuoka (identifying them by name via a programmable badge and keeping track of their movements with a built-in security camera), as well as a hospital porter bot that can carry luggage and guide people to elevators and a cute receptionist bot that converses in local dialects and prints out maps for visitors.

And, of course, there's Asimo, Honda's spacemanlike android who most recently made headlines for conducting the Detroit symphony orchestra. Asimo, the fulfillment of a two-decade project, can run at six kilometers per hour, kick a football, serve coffee and tea, yield the right of way to an oncoming pedestrian, respond to verbal commands and recognize faces. Honda hopes to have Asimo ready for commercial sale as a receptionist in five years.

Not to be outdone, last December Toyota president Katsuaki Watanabe declared robotics would be one of the company's core businesses within the next decade, and envisioned a society "where robots and humans live side by side." Toyota plans to double the number of its robotics engineers to 200 by 2010. Since 2004, it has showcased its prototype "partner robots," shiny white androids that can carry groceries and play "Moon River" on the violin and trumpet. Partner-robot bands have graced the stage at Washington, D.C.'s Kennedy Center and at the 2005 World's Fair in Japan's Aichi prefecture. The expo's droidfest also featured bots that could draw portraits, paint pottery, carry golf clubs, hit baseballs, sample and identify wine, and ballroom-dance. "I would not be surprised if I see robots taking the place of the elevator conductor in a Tokyo department store," says Masahiro Mori, a pioneer of Japanese robotics.

But the Achilles' heel of Japanese bots is cost effectiveness. Complex designs are expensive, and even the most successful Japanese bots have big price tags. Asimo leases (not very often) for $20,000 a day. Sony's $2,500 robotic dog, Aibo—unveiled in 1999—inspired a cultish following, but after selling 200,000 units, Sony put the dog down in 2006 (along with the rest of its robotics division) over concerns about the bottom line. "The Japanese have not been able to produce a mass-hit product because they're so enamored with the fantasy of creating a robot partner [that acts] like a human being," says Tim Hornyak, author of the robot-industry blog Loving the Machine. "They're just not satisfied with a hockey puck that will go around picking garbage up off your floor."

American firms do not share the burden of sci-fi fantasies. The pragmatic direction of their research is dictated in large part by the U.S. Department of Defense, which provides much of their funding. Roomba gets its ability to cover a surface thoroughly from mine-hunting algorithms developed for military bots. Similarly, Coroware's wheeled Corobot uses technology from U.S. Army unmanned vehicles. Since 2005, iRobot has parlayed this work into several single-purpose workhorse droids. Dirt Dog sweeps decks and garages, Scooba washes floors, Looj cleans gutters, each for under $500. The strategy has paid off: iRobot's on track to increase its $249 million in revenues by 20 percent.

Japanese and American firms have their eyes on the same prize: the market for home health care, particularly for the elderly. As baby boomers hit retirement age, the need to monitor and assist seniors will create a surge in demand for personal-care robots, experts say. Since 2001, the Japanese government has spent $210 million on research to meet its goal of deploying robots to support its aging workforce. (It's timeline specifies that bots should be able to straighten a room by the end of this year, make beds by 2013, and help with baths and meals by 2025.) The desire to field human-like robots, however, is an impediment. Honda, for instance, decided to keep its Asimo robot bipedal, even though its two feet are impractical in homes with stairs and clutter. The one field in which Japanese robots have a clear lead requires no practical applications: entertainment robots, a $185 million market that is expected to rise to $3 billion by 2014, according to private research firms.

The few Japanese droids that appear viable for the home-care market face one key problem: safety. Riken's Ri-Man, for example, has soft silicone skin, a wheeled base and two paddle-shaped arms strong enough to lift a child from a bed (the company is currently practicing the technology on dolls); its vision, hearing, smelling and touching sensors allow it to locate voices and respond to spoken commands. But at 1.5 meters tall and 99 kilograms large, Ri-Man could easily crush a child or senior if it accidentally tipped over.

Unlike Japan, the U.S. government has shown little interest in leading a national robotics strategy. Instead, private firms are partnering with university research centers and encouraging the development of software to guide home-care robots. The biggest player is Microsoft, followed by Willow Garage, a start-up founded in 2007 by early Google architect Scott Hassan. By 2009, Willow Garage plans to distribute up to 50 models of its PR2 robot (originally developed at Stanford) to university labs around the country, as a hardware-and-open-source software package, which researchers are free to modify. The PR2, lightweight and therefore safe, has wheels and grasping hands that will allow it to push a vacuum, dust cobwebs, hold open a door or perform other tasks.

Similarly, 50 or so research labs and private firms have adapted Microsoft's Robotics Studio software for use on their own bots. MIT's Media Lab is developing the Huggable, a robotic teddy bear that can be controlled remotely to transmit data about vital signs like blood pressure and heart rate, or to virtually embrace and chat with a faraway grandparent or child. The uBot-5, developed at the University of Massachusetts at Amherst, is compact enough to navigate narrow household corridors, and if a human bumps into it, it merely skids across the floor.

All this grass-roots robotics innovation has led tech giants like Bill Gates and Apple co-founder Steve Wozniak to predict that in the next twenty years, robots could be the biggest technological revolution since PCs and the Internet. Whether these robots are cleaning up homes or serving as co-workers, entertainers and friends depends on which vision wins out.