I, Robot

Subscribe to BusinessWeek

Nearly one million robots are chugging away around the world right now, doing work that is normally referred to, in robot circles, as "the three Ds" -- too dull, dirty, or dangerous to appeal to humans. But anticipating the day these machines take on white collar tasks, the great thinkers of the robotic world are cogitating a lot right now about kansei. It's a Japanese term that describes a group of perceptions and faculties currently beyond the reach of machines, such as intuition, pleasure and pain, curiosity, emotion, sensitivity, attachment, and creativity.



If robots are to gain widespread acceptance in the workplace, then they too must acquire at least the appearance of kansei, says Shuji Hashimoto, director of the Humanoid Robotics Institute at Waseda University in Tokyo. Otherwise, we humans, fed on a steady diet of evil robot movies, will shun the blinking, bleeping machine in the next cubicle -- or even turn it off.



Send in the Nursebot

For a rudimentary example of what an emotional robo-colleague might look like, meet Pearl, a lovely, red-lipped robot who resides at Carnegie Mellon University's Robotics Institute in Pittsburgh. Pearl, with a wheeled base and a smiling, metallic face, is a prototype "nursebot," built to care for the sick and the elderly in their homes. Health care is shaping up to be one of the biggest job markets for human-seeming robots, at least initially, because as populations of industrial societies around the world age, they are all projecting huge shortages of nurses and other health-care workers. Robots are naturals for picking up the slack, but only if the infirm can stand being around them.



When it comes to skills, Pearl is hardly unique. She can follow patients around and receive their commands through a touchscreen or by a spoken voice. And she can speak to them in plain English -- as long as the conversation doesn't stray from health. Plenty of other robots can handle similar chores. Aidu Chuo Hospital in the Fukushima region of Japan uses a guide robot to greet visitors and lead them to where they need to be; it looks something like R2-D2 from Star Wars.



What sets Pearl apart is that she has a humanoid face that's, well, pretty. A team of roboticists, social scientists, and software specialists, using theories drawn from cognitive science and the principles of aesthetics, created a face that her customers would enjoy interacting with. Pearl can smile, raise her eyebrows, and tilt her head, all in an effort to make patients feel at ease.



Designing Robots as Colleagues

Empathetic robots have not been much of a concern up to now. Who cared what a robot looked like when it was screening chemicals, painting Hondas, or exploring the toxic wreckage of Chernobyl? But the next generation of smart machines is meant to work beside us in offices, hospitals, and malls. "It has now become necessary for robots to interact on a higher level with humans to be of use," says Hashimoto. "Robots need to be tuned into human sensibilities."



That will be especially true for the kind of robots that have long captured the imagination of the public -- walking, talking 'bots with two legs, two arms, and a head that has some semblance of a face. This so-called humanoid architecture is not necessarily the most efficient robot design, but "the physical world is designed for humans. If robots are to exist in this world, they must move like humans," says David Bourne, a senior scientist at Carnegie Mellon's Robotics Institute.



That's not true at all of the Rube Goldberg-esque contraptions found in most factories. Human interaction with these robots is minimal, as factories often keep them segregated from living workers. The industry learned its painful lesson in 1981, after a worker at a Kawasaki Heavy Industries plant in Japan scaled a fence to fix a haywire robot. The machine, which could not sense his presence, shoved him to his death in a grinder -- possibly the first murder by robot.



Asimov's Laws

Consequently, roboticists have long embraced "first, do no harm" as the overriding directive of man-machine relations. Their guiding principles come from famed science fiction writer Isaac Asimov's three laws of robotics, first formulated in his 1942 short story, Runaround: 1] A robot may not injure a human being or, through inaction, allow a human being to come to harm; 2] A robot must obey orders given to it by human beings except where such orders would conflict with the First Law; and 3] A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.



In the best of android workplaces, a robot would figure out on its own when it is safe to turn on its superhuman strength and when to be gentle. Robot developers around the world are trying to design machines that can sense when people are around and behave accordingly. This might mean the 'bots keep their distance when they must and return to their chores once the human is safely out of the way.



That kind of sensitivity requires highly complex programming task -- and there is lots of work ahead. So far, roboticists have barely been able to develop humanoids with the learning capacity of a none-too-gifted toddler. Reading human reactions will probably take another decade of research, at a minimum.



Robotic Small Talk

One approach under development is to cover humanoids with sensors that can pick up physiological cues such as heart rate, facial muscle movements, and sweat; and can adjust their actions accordingly. Yet another level of programming complexity is added to coax the face of a machine like Pearl to express fear, happiness, or puzzlement.



Keep in mind, though, that robots really don't have to be as sophisticated as humans to do their jobs. "Robots only need to give people the perception that there is some awareness there," says Michita Imai, associate professor of the human-robot interaction lab at Keio University in Japan. He believes that the big near-term challenge for robot designers is simply to spark interaction with humans.



"We have one experiment," he says, "where a robot offers candy to a study subject. After the person eats it, the robot says, 'Yummy isn't it?' If you repeat that pattern of having the robot prompt a person's response, the person begins to assume that the robot can relate on a certain level. It makes the person more likely to listen to what the robot is saying." Proof that a little bit of kansei goes a long way with humans.



For more, visit BusinessWeek's slide show.