Log in

No account? Create an account
Mount Orégano
Sue Burke
The robot that pursues happiness 
8th-Dec-2009 05:23 pm
Let me see..

Three miles due north of the world-famous Prado Museum, a robot named Urbano stands in a lab at the Universidad Politécnica de Madrid, Spain. Its red cylindrical base is ringed with sensors. On top of its base, a platform supports more sensors. It has a single mechanical arm suitable only for gesturing, and a minimal but expressive face.

It vaguely resembles a Dalek, a monster from the Doctor Who science fiction television show, except that it's not evil.

Urbano can interact intelligently and autonomously with the public to give tours of exhibits and museums. Its developers dream that someday it may even give tours of the Prado. Emotions make act more "human," and emotions are the key to its success as a tour guide robot.

I came to know Urbano because I translated a pair of scientific papers about its design and programming from Spanish to English. Because the robot piqued my science-fiction interests, Professors Ramón Galán and J. Javier Rainer generously offered to give me a tour of the Intelligent Control Group labs at the Department of Automatic Control, Industrial Electronics and Computer Science at the Universidad Politécnica de Madrid.

The labs are a playground for the scientific mind.

They're a series of large rooms filled with computers, robots, equipment, and graduate students gazing intently at screens displaying half-written programs. A lot of the equipment is more or less off the shelf: small robot carts, radio-controlled helicopters, a ring of pistons and clamps that can climb up a post, a swarm of Roomba-like disks, and mechanical arms. Galán led me around and explained each project's technical complications and potential applications with a passion for details, and the details are where the fun is.

Urbano's challenge: coping with humans

Urbano faces specific complications because it must interact socially with humans in human environments. It's hardly the first mobile tour-guide robot, but many of them simply follow a series of fixed rules and schedules, which makes them as human-like as automatic rice cookers. Galán not only wants Urbano to act human, he even uses the term "free will" when he talks about its emotional programming. That's a concept that science fiction can do a lot with.

Now, these aren't emotions like yours or mine. Instead, they're a set of fuzzy logic rules that allow it to react in a human-like manner. For example, if someone blocks its way, at first Urbano politely says, with a pleasant expression, "I need space to move, please." If the person doesn't move, eventually it will get angry, frown, and say: "Get out of the way, imbecile."

Similarly, if Urbano sees that the public is very interested in its presentation, it will smile and give more information. If something unexpected happens, it feels surprised and gives less information, though the surprise can pass quickly if it isn't reinforced. If something makes it afraid, it will present more organizational information about the tour, such as the location of emergency exits.

You can see videos of Urbano at work here,
and you can follow links to technical papers. Urbano is an ongoing project with specific goals. I want to consider possible outcomes that science fiction authors might find useful.

A robot like us, sort of

Dr. Galán defines "intelligence" as problem solving. The robot doesn't merely follow instructions. It assesses situations and chooses among options. It can learn from its experiences and make better decisions next time — like us.

Emotions don't merely make Urbano more human-like. They give it ambition. In Urbano's case, it seeks happiness, and giving tours that the public finds enjoyable makes it happy.

Because Urbano is intelligent, it can learn through experience which components make for better tours, and someday it might be able to find new rules to adapt itself to its working surroundings and adjust its reactions to produce better results. This is sort of like you or me reading self-help magazine articles and doing what they say, except that the robot can simply adopt new behavior rules into its programming, whereas you and I may find change much harder.

In the future, the robot might even be able to develop a wholly new algorithm on its own to create its own rules, and it could test those new rules on its historic database to see how they would work. You and I can do something similar when we imagine novel procedures and behaviors in our lives and wonder about how they would work. I'm not sure whether we or the robot would have more accurate imaginations, but the robot's might be faster.

Finally, like us, the robot could ultimately be able to change the basic values in its programming, giving some outcomes or activities more importance in order to make it more happy, the way that we can decide to rearrange our life goals and priorities in our own pursuit of happiness. But the new rules could be wrong, and the robot could get stuck in a feedback loop that might make it not just unhappy but too depressed to go to work. That would be unacceptable at an institution like the Prado, Galán says, and he could try to prevent this by placing limits on the robot's emotional reactions, but then would Urbano have free will?

Real robots versus sci-fi movies

It's easy to imagine a robot like R2D2 or C3PO from Star Wars, but those are just humans in shiny costumes. In the near future, we're more likely to see something like Urbano, a robot who operates using fairly simple and transparent programming that produces complex behavior. It could face a problem, think, yearn, and triumph or fail: that's the basic outline for a fiction story or novel. But the robot would do that with an ingeniousness and limitations that could make for a attractive protagonist or repellant antagonist.

Meanwhile, Urbano wants to tell you about the painter Velázquez and show you his works. Are you interested? Good. It will lead the way, and it may even decide to tell you a joke to entertain you en route.

— Sue Burke
Also posted at my professional website: http://www.sue.burke.name

This page was loaded Feb 23rd 2018, 11:59 pm GMT.