杏吧原创

ROBOTIC helpers are not yet in every home. But in recent years robots have steadily marched into the real world to perform tasks such as cleaning floors, delivering drugs or simply entertaining.

That has let anthropologists and roboticists give these mechanical workers their first report cards 鈥� and results are mixed. Despite evidence that we can find robots useful, even lovable colleagues, they can also trigger annoyance and violence. The results should help make future robots easier to work with.

One study by at Carnegie Mellon University in Pittsburgh, Pennsylvania, highlights how popular culture can affect a robot鈥檚 reception.

People she introduced to Roomba, a robotic vacuum cleaner made by of Bedford, Massachusetts, compared it with their knowledge of robots that explore Mars, forming low expectations of Roomba鈥檚 abilities. But making a bad first impression seemed to help Roomba; it invariably surpassed expectations, helping people bond with their machine.

A six-month study of how Roomba affected households, conducted by at the Georgia Institute of Technology in Atlanta, backs up that finding.

鈥淪ome people saw it as a lifetime partner 鈥� they had a real emotional attachment to it.鈥� Even those who returned to their previous cleaning routine didn鈥檛 blame the robot, instead saying it was their routine that was at fault.

Yet not all robots have fared so well. A handful of hospitals across the US use a robot called TUG, made by Pittsburgh-based firm , to carry drugs between wards.

When Bilge Mutlu at the University of Wisconsin, Madison, studied how hospital workers got along with TUG, which navigates hallways and uses elevators, he found it polarised opinion. Some saw it as a team player and loved it; others thought it was attention-seeking and resented it (see 鈥淟ove, hate and a hospital robot鈥�).

These studies may be small scale, but they help illuminate the skills robots need if they are to get along with humans. The designers of the few robots to have reached the real world and work alongside people are breaking new ground, just not always successfully.

Maria H氓kansson at the Swedish Institute of Computer Science in Stockholm recently ran a 10-month project to see how six families got on with the , made by Innvo Labs in Nevada.

Pleo鈥檚 impressive behaviours, like responding to sounds or touch, led to initially positive reactions. But niggles like battery life, and a perception that it wasn鈥檛 actually as capable of developing a personality as claimed on the box undermined those impressions, says H氓kansson. Frustrated by seeing no evidence of learning, one person declared: 鈥淧leo was just as stupid as a week ago.鈥�

Robots that come across as sophisticated and smart, unlike Roomba, are in trouble if they are seen not to deliver.

Back in the lab, some roboticists think they have a solution, says Kathleen Richardson, an anthropologist studying robots at University College London. 鈥淸Researchers] design lots of robots to look like children so that people will imagine they have a more childlike mentality,鈥� she says. In this way they attempt to guide those crucial first impressions and induce people to be more forgiving of mistakes.

Similarly, it has also been assumed that a functional appearance affords protection from high expectations and later disappointment, something that studies of Roomba back up.

Still, Mutlu鈥檚 work shows a humble appearance doesn鈥檛 always have that effect. Because the box-like TUG robot could navigate independently and used a human-like voice, 鈥減eople expected it to show appropriate behaviours related to these qualities,鈥� Mutlu says.

A box on wheels it may be, but TUG was also a hospital worker, and its colleagues expected it to have some social smarts, the absence of which led to frustration 鈥� for example, when it always spoke in the same way in both quiet and busy situations.

Those findings aren鈥檛 going unnoticed: Aethon is looking into some of the issues raised by the TUG study. 鈥淲e have redesigned the cart and user interface controls, adjusted voice volumes and included the ability to make voices louder or softer at particular times of day,鈥� says Dave Wolfe, product manager at Aethon.

As more studies like these feed back to the designers of robots to work alongside humans, their products should find fewer kicks aimed in their direction. For many of those on the market now, though, the report card still reads: could do better.

鈥淚鈥檓 on the phone! If you say 鈥榬obot has arrived鈥� one more time I鈥檓 going to kick you in your camera鈥�