Visit Your Local PBS Station PBS Home PBS Home Programs A-Z TV Schedules Watch Video Donate Shop PBS Search PBS
SAF Archives  search ask the scientists in the classroom cool science
scientists from previous shows
cool careers in science
ask the scientists

Photo of Illah R. Nourbakhsh Illah R. Nourbakhsh as seen on Natural Born Robots: Robots Have Feelings, Too

Click on Illah's photo to read a brief bio.

q What are the "emotions" Sage records and reports, and why did you pick these specific feelings? (Question sent by Susie)

A Sage was designed so that we can easily plug in and pull out different "emotions." In fact, after the overall architecture was created and tested, we sat down with the Education Division at the Carnegie Museum and we asked the educators there to choose the right emotions for Sage. The critical point was to choose emotions that are relevant -- because they can be used to inform the robot's real-time behavior and because they are the emotions most obviously affected by its interaction with people.

We implemented all of the emotions that the Education Division suggested:


Now, you might argue that being tired is not an emotional state, but we define emotions very broadly. Each of these four is affected by the robot's environment, and each one of these has the potential to trigger the robot to behave differently. So, for example, when Sage's batteries are running low, it gradually becomes more and more tired, only grudgingly giving tours and soon returning to plug itself into the wall. If museum visitors pay attention to its video clips, Sage becomes happier and happier, making its happiness clear by telling many jokes and generally acting hyperactive. Leave Sage alone for longer and longer and it gradually becomes lonely. Once it is lonely, it will go out on a limb to attract a museum visitor and give them a tour. It will even make people feel sorry for it! Stand in Sage's way and happiness can gradually turn to frustration as it tries to convince you to let it go on its way. Stand in Sage's way when it's lonely and, instead of becoming frustrated, it actually becomes happy to see you!

Why these "emotions?" Because these reflect how people and the robot are interacting really well, and so these are the emotions that really matter to a tour guide that hopes to give exciting tours. The important thing is for Sage to wear its emotions on its sleeve -- to respond transparently to the stimuli which humans provide it.

Now, you can read more about the Sage system by downloading a PDF version of a paper that describes it in great detail. You can get to that at by downloading the document, sage.pdf.

Now, an interesting update is that we've started doing emotional robots differently recently. After talking to a drama teacher, we learned that great actors don't think about emotion, they think about action. It is through their action that the audience infers emotion. So, we're making our robots more like actors and less like real humans, and it works really well! The project is called Robot Comedy Improv, because that's what the robots do, and you can read about this new work on affective robotics by downloading the paper, robotimprov.pdf from the same website.

q How do you respond to the e-mails Sage sends you at the end of the day? If it reports a really bad day and a lot of frustration, do you reprogram it so it starts from a neutral position the next day? (Question sent by David P.)

A Well, the most important emails are those that ask for urgent help. It looks like a scene out of an Emergency Room when we get one of those, since we race down to the airport, notebook computer in hand, and check Sage's vital signs. It's actually neat that, in situations like that, life support matters, just like it does for humans. Sage runs on battery power, and if something goes wrong that keeps him from taking care of himself, he can run his batteries down to the point that they are damaged. So, there really is a sense of urgency when these e-mails arrive. They are, happily, quite rare.

Sometimes, Sage sends us email that tells us that he is having trouble plugging into the wall, or that he is having trouble navigating part of his route. In cases like this, we go down to the museum and visually check the environment to see if something has changed.

In one case, Sage was becoming frustrated very often for about a week. Then we received a phone call from the museum guards, stating that Sage was going down the halls somewhat "drunken." By the time we identified the problem, it turns out the alignment on the wheels had been ruined by a faulty sensor, and the result was the Sage was still functioning, but drunk!

q I would like to see a robot guide in action. Is the Carnegie Museum the only museum that has one? If not, where are the other guides and are they a lot like Sage? (Question sent by several viewers)

A In fact, there are currently three robots with our technology giving tours! There is a spin-off company called Mobot, Inc. ( that makes them and sells them now. Two are in the Carnegie Museum of Natural History. A second robot, upstairs, is a *she*, and she is a tour guide in a safari of North American mammals.

The third robot, called Joe, is a really funny Pittsburgh robot with an attitude. He teaches people how to speak with a Pittsburgh accent at the Heinz History Center in downtown Pittsburgh. Pretty soon, I bet Mobot will have more of these robots installed in other cities as well!

q I think it would be neat if a robot with feelings could be a companion to people who are shut-ins or lonely -- singing to them and providing company. Do you think this is possible? (Question sent by Jeluo, Freshman at Homestead High School)

A Absolutely, yes. In fact, this is a great way for robots to help provide companionship while still letting people feel like they are in control of their lives -- often when you give the elderly a full-time caregiver, they feel like they are losing autonomy, and so a robot can act like a very good companion. Singing and playing games should be no problem for a robot. Playing games will especially benefit from advances in computer vision that are taking place every day now. I expect card-playing robots are just around the corner now.

Here at Carnegie Mellon's Robotics Institute, there is even a project called NurseBot in which Sebastian Thrun and his colleagues are designing just this -- a tall, moving, entertaining robot that can keep someone company in their own home. You can learn more about this project by going to the web page:

q What robot is the most advanced "feeling-wise" that is either being built now or has already been built? (Question sent by Jewelgrl)

A That is a difficult question to answer. We have two robots that can do comedy improvisation -- so they are very spontaneous and susceptible to reacting emotionally to each other's behavior. Chips is certainly very emotional in the sense that it actively tracks its feelings. Kismet at MIT is also quite emotional. But, you know, in the grand scheme of things, none of the robots that exist today are very advanced in terms of "feelings," even compared to a four-year-old child. So, I would say robotics has a long way to go until we can truly say that our robots have feelings!


Scientific American Frontiers
Fall 1990 to Spring 2000
Sponsored by GTE Corporation,
now a part of Verizon Communications Inc.