Saturday, February 27, 2010
The culmination of two years of research and the collective expertise of 17 faculty members, undergraduates and doctoral students in the Human Robot Interaction Group, it is a robot outfitted with a $20,000 laser navigation system, sonar sensors and a Point Grey Bumblebee 2 stereo camera that functions as its eyes, which stare out from its clay-colored plastic, gender-neutral face.
With Dr. Rybski looking on like a proud parent, a bearded graduate student clacked away at a laptop on a roving service cart, and the robot rolled forward to fulfill its primary function: the delivery of one foil-wrapped Nature Valley trail-mix flavor granola bar.
“Hello, I’m the Snackbot,” it said in a voice not unlike that of HAL 9000, from “2001: A Space Odyssey,” as its rectangular LED “mouth” pulsated to form the words. “I’ve come to deliver snacks to Ian. Is Ian here?”
I responded affirmatively. “Oh, hello, Ian,” it said. “Here is your order. I believe it was a granola bar, right?”
Yes, it was. “All right, go ahead and take your snack. I’m sure it would be good, but I wouldn’t know. I prefer a snack of electricity.”
Designed to gather information on how robots interact with people (and how to improve homo-robo relations), the Snackbot has been carefully considered for maximum approachability in every detail, from its height to its color. The snack, not surprisingly, is the central component of that approachability.
“We figured, what better way to get people to interact with a robot than have something that offers them food?” Dr. Rybski said.
The Snackbot is but one soldier in a veritable army of new robots designed to serve and cook food and, in the process, act as good-will ambassadors, and salesmen, for a more automated future.
In 2006, after four years of research and more than a quarter-million-dollar investment, Fanxing Science and Technology, a company in Shenzhen, China, unveiled what was called the “world’s first cooking robot” — AIC-AI Cooking Robot — able, at the touch of a button, to fry, bake, boil and steam its way through thousands of Chinese delicacies from at least three culinary regions.
AIC-AI needs a special stove for cooking, but many of the mechanized culinary wizards developed since then can work on almost any kind of stove, as long as the robot is either shown ahead of time how a particular stove works or the stove’s characteristics are programmed into the robot’s software.
In 2008, scientists at the Learning Algorithms and Systems Laboratory in Lausanne, Switzerland, came out with one such teachable chef, the Chief Cook Robot, which can make omelets (ham and Gruyère were in its first) and bears a resemblance to the Pillsbury Doughboy. That same year, at the Osaka Museum of Creative Industries in Japan, a programmable robot began preparing takoyaki (octopus balls) from scratch, a chef’s bandana wrapped jauntily around its upper module.
Then, a month later in Nagoya, Japan, the Famen restaurant opened, with two giant yellow robot arms preparing up to 800 bowls of ramen a day. When it’s slow, the robots act out a scripted comedy routine and spar with knives.
“The concept of this restaurant is that Robot No. 1 is the manager, which boils the noodles, and Robot No. 2 is the deputy manager, which prepares for soup and puts toppings,” said Famen’s owner, Kenji Nagaya. “Human staffs are working for the two robots.”
In the throes of an economic downturn, with unemployment rates mounting, the very idea of a robot chef might seem indulgent at best — at worst, downright offensive. But these robots aren’t likely to be running the grill stations or bringing you chowder anytime soon — and the bad economy might be part of the reason. At $100,000 a pair, Mr. Nagaya said, the cost of his robots is “too high to make bowls of ramen.”
Mikio Shimizu, the president of Squse, a company in Kyoto, Japan, that is responsible for the sushi-grabbing hand, said that his ultimate goal is to become the world’s largest maker of functional prosthetic hands.
Narito Hosomi, the president of Toyo Riki, a company in Osaka, Japan, that programs the robots responsible for the octopus balls and savory pancakes, said that the final destination for the robots, which cost $200,000 each, was more likely a factory than a kitchen.
But “it’s not interesting to watch robots welding,” Dr. Hosomi said. “If you see robots do the same work as you do in everyday life with the tools you use, it would be easier to understand the functional capability of robots. The okonomiyaki robot is a medium for that purpose. We say a robot can make okonomiyaki, takoyaki — well, what would you like a robot to do for you?”
While cooking is certainly a more universal way to showcase a robot’s abilities than, say, laser-welding, it is also unique in its ability to tackle something deeper: namely, the public’s collective “Terminator”- fueled angst over a future populated by vengeful hominoid machines.
Dr. Heather Knight, a roboticist at the NASA Jet Propulsion Laboratory, said that the industry is trying to change “the perception of robots.”
“The Japanese have always been more comfortable with it, but particularly in the West, there’s this whole Frankenstein thing that if we try to make something in the image of man, to make a new creature, we’re stealing the role of God, and it’s going to turn out wrong because that’s not our role,” she said. “So how do you change this perception that robots are going to be way too intelligent and destroy us? One of the fastest ways to people’s hearts is food, right? Any girlfriend or wife would say that.”
In fact, Dr. Aude Billard, whose team designed the egg-handling Chief Cook Robot, said that she decided on omelets because “it was the first dish my partner cooked for me.” The omelet making was meant to show how a robot could be “taught” to accomplish complex tasks. It was also “something that all the guys in the lab knew sufficiently well to be able to train the robot,” she said.
But perhaps the biggest accomplishment of this new wave of sustenance-bearing machines is their departure from what defined their predecessors. The Fritz Lang level of efficiency normally associated with robots is notably absent — and that’s no accident.
“A simple rule of robotic personality seems to be: don’t make things the most efficient way,” said Magnus Wurzer, who has been running the Vienna-based Roboexotica, a festival where scientists have gone to build, showcase and discuss “cocktail robots” since 1999.
One entry, Beerbot, detects approaching people and asks for beer money. When it acquires enough, it “buys” itself a beer. Bystanders can watch it flow into a transparent bladder. As for other humanizing behaviors, “like a robot that doesn’t stop short at lighting a cigarette but actually goes ahead and smokes it?” Mr. Wurzer says, “We had that.”
Roboexotica has inspired a stateside version as well, which just had its third annual celebration in San Francisco.
And in at least one case in Europe, a robot actually got behind a bar. From 1999 to 2002, a scarlet-eyed metal robot named Cynthia poured drinks at Cynthia’s Bridge Bar and Lounge in London. But according to Mr. Wurzer, “she was too costly to maintain once the bar was sold by the robot’s maker.”
One reviewer at virtual-london.com, a travel-information Web site, said that Cynthia’s problems went deeper: “She whirls into action, pouring drinks to perfection, mixing them, recounting awful jokes and chuckling to herself while frightened customers feel grateful she’s not allowed out from behind the bar.”
However hard it may be to master, humanizing behavior was what the Snackbot’s creator were seeking, too.
“How do you get a service robot to interact with humans?” Dr. Rybski asked. “That’s a real hard problem. It’s different when you’re working with a human versus a pipe on an assembly line.”
To prepare, one of Dr. Rybski’s graduate students, a slender and quiet Korean woman named Min Kyung Lee, spent two days staked out behind a campus hot dog vendor, taking notes on how he interacted with his customers. She used what she learned to program the robot’s dialogue.
Aside from the obvious challenges of instilling a machine with personality is the other, long-held axiom in the world of robotics: what might seem second-nature to humans can be all but impossible to teach a machine.
Mr. Wurzer said that one scientist at Roboexotica built a robot solely dedicated to the preparation of mojitos — “with the grinding and stomping and all.” And yet the most challenging task for all the robots, he said, was probably the one thing that no human bartender ever botches: handling the ice.
The Chief Cook Robot still relies on human beings to crack the eggs — the shells are far too delicate for its metal hands. The okonomiyaki-making robot still needs the vegetables prepped, a task arguably better suited to a robot.
And while robots could certainly be developed and trained for these tasks, some culinary arts are so delicate and ancient — so venerated and sanctified — that even these machines’ creators wouldn’t trust them to inhuman hands.
“Would you like to have a robot hand that makes sushi?” said Mr. Shimizu, of Squse, which programmed the sushi-grabbing hand. “Do you really want it? For making good sushi, a robot never can beat a human professional sushi chef. A robot never can go beyond a human’s skill or human intelligence.”
Keizo Shimamoto, who writes a blog on ramen noodles and has eaten at Famen, the two-robot Japanese restaurant, said that the establishment was “kind of dead” when he ate there last year. Though the owner said that people do taste the food, according to Mr. Shimamoto, “It was a little disappointing.”
It’s one thing to get people to stop by to see the robots. “But to keep the customers coming back,” he said, “you need better soup.”