Tech industry looks to robots to tackle problems but finds hurdles on the way

Robotic devices are everywhere: in factories, law enforcement, caretaking. They even suck up dust bunnies. Today they are smarter than ever, but they only excel when the task is clearly defined. Science correspondent Miles O'Brien reports on why it's hard to teach robots basic human things, like walking and problem solving.

Listen to the Audio

JUDY WOODRUFF:  The latest gadgets and electronics are perennially best-selling holiday gifts. 

Our science correspondent, Miles O'Brien, had a more ambitious wish list, but he didn't yet get what he wanted for Christmas. 

ACTOR:  Danger, danger.  Force level building to fatal intensity.

MILES O'BRIEN:  I have always wanted what Will Robinson, George Jetson, and Luke Skywalker have.

ANTHONY DANIELS:  That's how he came to be in your service, if you take my meaning, sir. 

MILES O'BRIEN:  You know, a robot servant to do my bidding, my dirty work. 

Seems like that idea is languishing in around-the-corner purgatory, with the flying car, fusion power and the jetpack.  A lot has changed over the years.  Robotic devices are everywhere, assembly lines, disarming bombs, helping the disabled, even sucking up dust bunnies.  They are smarter than ever.

But, unlike Hollywood's robots, they only excel when the task is very narrow and clearly defined. 

PIETER ABBEEL, University of California, Berkeley:  A robot is really easy to fix to do a repeated motion.  The science, it's very hard to deal with variation like this when, every time, it is looking at the towel and seeing something different. 

MILES O'BRIEN:  This really hit home for me when I saw this robot in Pieter Abbeel's lab at U.C. Berkeley. 

I'm going to out-fold this robot, darn it. 

He has taught the device to fold clothing.  In the world of robots, that's a big deal.  It takes about 20 minutes to fold one measly towel.  Why?  Computers are smart enough to beat the world's greatest chess master.  Why are robots flummoxed by a dirty rag?  Well, it's complicated. 

MARVIN MINSKY, Cognitive Scientist:  There's still no machine that can solve everyday commonsensical problems. 

MILES O'BRIEN:  Marvin Minsky helped create the field we call artificial intelligence, you know, making computers think like us.  Over the years, he has stumbled on a surprising paradox:  What's hard for us is simple for robots, and vice versa. 

MARVIN MINSKY:  If somebody is very good at some skill, it's because they know about 20,000 fragments of knowledge or process or whatever. 

MILES O'BRIEN:  But to have common sense, the mundane skills you need to get through the day and fold the clothes, you need a few million fragments of skill, knowledge and insight. 

MARVIN MINSKY:  So, this advanced mathematics came easily, and then the high school-type mathematics was a little later, and we are still not at the age of the 4- or 5-year-old. 

MILES O'BRIEN:  Humanoid robots are also having a hard time learning to walk. 

How difficult a problem is it? 

ROBERT PLAYTER, Boston Dynamics:  It's it's difficult because we don't know what we don't know. 

MILES O'BRIEN:  Rob Playter is with a company called Boston Dynamics.  This is the home of Big Dog.  Big Dog was built at the behest of the Pentagon's Defense Advanced Research Projects Agency.  They're seeking a mechanical mule for foot soldiers in the next four years or so. 

Powered by a two-stroke engine, hydraulic actuators, a gyroscope, and some breakthrough software, this prototype can slog through some pretty rugged terrain.  It has a remarkable sense of balance and the ability and agility to break a fall. 

So, what do you do?  What is the way to -- how do you -- how do you teach a machine to walk? 

ROBERT PLAYTER:  Well, you have to -- you have to build them.  You have to experiment with them.  You have to push them.  You have to kick them and see how they respond.  Rather than try to build a response to stepping on a rock or stepping on ice, what we try to build is a fundamental sort of core concept of balance and how to behave in -- in the gravitational field. 

MILES O'BRIEN:  Boston Dynamics is now developing at a two-legged robot called PETMAN for the Army. 

And, in Florida, at the Institute for Human and Machine Cognition, they're working on some legs with a keen sense of balance that may one day be connected to Robonaut.  But to get to this point, well, let's just say it's been a long, slow stroll. 

MARVIN MINSKY:  What most people today are doing is saying, first let's get the robot so that it can do the simple things, and then we will make it do the harder ones.  I think we should just turn it opposite.  

MILES O'BRIEN:  While others try to solve the ambulation equation, at MIT's Media Lab, Cynthia Breazeal is focused on this question:

CYNTHIA BREAZEAL, Massachusetts Institute of Technology Media Lab:  Should it be a human device?  How humanlike should it be?  What do we even mean when we say ‘humanlike’? 

COMPUTER VOICE:  My name's Nexi.  What's your name? 

MILES O'BRIEN:  Nexi is just the latest robot in her menagerie that Breazeal has programmed to engender trust by bridging the gap between machine and mankind with expressions, nonverbal communication, body language, if you will. 

CYNTHIA BREAZEAL:  So things like when I finish speaking and I look at you, that's a very implicit prompt that now I'm expecting you to respond. 

MILES O'BRIEN:  And this is when the kids go wild, right? 

Breazeal is also fascinated with ways to make robots a better learning tool for children.  The more expressive and empathetic and frankly, cute the face is, the better. 

Now, here is a face anybody could love.  Looking at Leonardo, it is easy to forget what is behind him. 

SHERRY TURKLE, Massachusetts Institute of Technology:  Here are my -- all my little robots. 

MILES O'BRIEN:  So these are robots you like? 

SHERRY TURKLE:  Yes.  Well, you know, I'm obsessed with robots. 

MILES O'BRIEN:  Really? 

Sherry Turkle is a colleague of Breazeal's at MIT.  Her latest book is "Alone Together: Why We Expect More From Technology and Less From Each Other."  She says humanoid robot builders are leading us down a slippery slope. 

SHERRY TURKLE:  The moment you make a robot in human form, and the moment it can make eye contact, track your motion and gesture toward you, you're kind of toast, because you believe that there is somebody home, in other words, a consciousness, even potentially something with feeling and that is like you. 

MILES O'BRIEN:  But it isn't, of course.  Turkle worries about another paradox:  Machines that act like humans can dehumanize the real thing. 

SHERRY TURKLE:  And a lot of the fantasies about nanny-bots and elder-care-bots are really about being company, being companions for people who, quite frankly, we think sometimes we don't have time for.  And there, I think we get into a lot of trouble, because, you know, why are we doing this? 

MILES O'BRIEN:  So, this is love's labor lost? 

SHERRY TURKLE:  Love's labor lost.  It diminishes us.  It diminishes us as people. 

CYNTHIA BREAZEAL:  Now, I have faith that people are actually pretty savvy about relationships.  And the relationship I have with various people, whether it's my children or my husband or my parents, these are all of very, very different kinds of relationships. 

The relationship I have with pets is a very different kind of relationship.  I think people are pretty savvy. 

MILES O'BRIEN:  So when am I going to get my robot butler?  The makers of Big Dog say all the pieces are finally coming together: intelligence, expressions, dexterity, and mobility. 

JUDY WOODRUFF:  Since Miles first filed that piece, Google bought Boston Dynamics, the maker of Big Dog, in December.  It was the eighth robotic company the tech giant purchased in the past year.