A.When Takanori Shibata began working on robots in the early 1990s, he had somethingpractical in mind, perhaps to help the elderly with their daily chores. But he soonrealized that robots were not really able to do anything useful, so he decided to makea robot that did not even try--but that could nevertheless deliver real benefits.
B. He created and developed his robot--Paro since 1998. It is 57cm long and looks like ababy harp seal. Thanks to an array of sub-skin sensors (传感器) , it responds kindlyto stroking. And though it cannot walk, it can turn its head at the sound of a humanvoice and tell one voice from another. It is a comforting and gentle presence besideyou. The best thing about it is that it seems to be helping in the care of people withdementia ( 痴呆症 ) and other health problems.
C.You could see Paro as a very well-designed $5,000 pet, but it is so sweet that it willnever offend you. It is as happy in one lap as the next, needs no house-training, canbe easily washed and will not die. This makes it a much more practical suggestion tohave in a nursing home or hospital than a live pet. It is used in such homes in Japan,in parts of Europe and in America. As well as simply making people happy--no meangoal--it can act as a source of reassurance and calm. People with Alzheimer's oftensuffer from "sundowning" ( 黄昏症 ) --a distressed urge to wander that comes ontowards the end of the afternoon. Mr. Shibata has found that a seal in the arms tendsto reduce such wandering, which means fewer falls. Experience in some westerncountries indicates that care homes equipped with Paro need less medication fortheir residents. Larger trials to test the medical benefits and other benefits that can bebrought by Paro are now under way in Australia.
D.If Paro proves to be more useful than a lovely animal, there will be a huge marketfor it. Akifumi Kitashima, who works on Japan's robotics strategy, points out that in2025 Japan will have 10.7m more elderly people than it did in 2005. Though Japan isageing particularly quickly, a lot of the rest of the world is on a similar course. Somewill long remain agile; most will eventually need care.
E.Looking after old people in homes might become easier with robots, be they moodenhancers like Paro or something more practical that can help care workers liftand reposition their charges (Mr. Kitashima says 70% of carers have bad backs).Yoshiyuki Sankai, perhaps Japan's best-known robotics entrepreneur, has set up acompany called Cyberdyne to make wearable systems that help people walk and liftthings by adding artificial strength to their limbs.
F. Robots may also make it possible for old people to stay independent in their homesfor longer. Mr. Angle says this is iRobot's "long-term guiding star", towards whichthe Roomba is a small step. In the view of Mr. Gupta, the general-purpose home-helprobots would be a big advance which, given a push, could be achieved in a couple ofdecades. Mr. Thrun reckons it could be done more quickly. Mr. Ng points out that ifyou get a graduate student to teleoperate ( 远程操作) a PR2 robot, it can alreadydo more or less everything a home-help robot might be required to do, so all that isneeded is better software and more processing power, both of which are becomingever more easily available.
G. Cloud robotics ( 云端机器人技术 ) can probably provide much of the requiredsoftware. Mr. Pratt says that if there were dramatic performance improvements in thefinals of the DRC, he would expect them to come from the cloud. But specific robothardware will need upgrading, too. No robot hand yet comes close to the utility of thehuman hand. Tasks that require feedback in terms of force and fit--like putting a pluginto a socket--remain particularly hard for robots. General technological progresswill not help; the only way to find a solution to this sort of problem is to workspecifically on it.
 Even more important will be interfaces ( 界面 ) to tell the robots what to do. Take-me-by-the-wrist Baxter, stroke-me Paro and the film-enabling industrial arms of Bot& Dolly, all very different from each other, show that interfaces can matter just asmuch as any other technological advance. Tobias Kirmebrew, of Bot & Dolly, thinksthat new interfaces could open up markets and applications of robotics in all sorts offields, and might do so surprisingly quickly.
I.Voice would be an obvious choice, but it has its drawbacks: give a robot a voice, saysMr. Hassan, and the user will think it is smart. An interface that allows the robot tobe dumb and the user not to care might be preferable. Indeed, small errors can beendearing, and needing help with something can produce affection. People do notresent Paro's need to be stroked; it is one of the things they like about it. CoBot'sneed for help with the lifts at Carnegie Mellon makes people warm to it, though beingtroubled for help by random robots in offices and shopping malls would probably notwork so well. But if the interface is properly designed, teaching a robot home help todo the job better might make it more welcome.
J.It may also be a good idea to let the robots turn for help to people other than thosethey are working for. As Mr. Goldberg points out, the cloud does not just containcomputers; it provides access to a lot of humans, too. One of the things that makeAethon's Tugs a success in hospitals is that the company's headquarters has a smallbut always staffed help desk which deals with questions from robots. If one gets stuckor lost, a remote operator can look through its eyes, check its logs and sort things outbefore the hospital concerned even becomes aware that anything is wrong.
K. If the robot can call on a help desk, it can communicate with other people too. Somehome-automation products already allow a degree of monitoring, notes on Chambersof Carnegie Mellon's Quality of Life Technology Centre, but what they offer leavesmuch to be desired. It makes the adult offspring feel greater responsibility--whichthey often cannot exercise--rather than giving them reassurance. The elderly, for theirpart, can feel snooped ( 窥听 ) upon. A robot with a defined presence in the housemight make a better intermediary.
L. What matters, as iRobot and other practically minded companies have learned, is notso much having robots but having a business model that does a job, be it washing thedishes, checking that medication is being taken or providing telepresence ( 远程监控). Producing something reliable and likeable that can be sold in large numbers anddoes not get its makers sued may prove a lot more difficult than simply developingthe required robotic skills, but not impossible. To be sure, robots will not spread asquickly as mobile phones have done. Over a decade they may not achieve much. Overa century, though, they could turn everyday life upside down.