How “Social Robots” That help Kids With Autism Work

When we think of robots, many of us dream of having at home a Robotic, the Cybernetic Assistant of The Supersonic, that cleans, cooks, washes and irons. But the truth is that today we have artificial intelligence in the palm of our hand: cell phones, voice-controlled digital assistants, and smart vacuum cleaners, like Rumba, who can once get trapped in a carpet, but clean without needing a human nearby.

Another type of robots are those that manufacture at Yale University,which improve the social skills of children with autism through work similar to that done by therapists and with a particularity: they manage to teach and express emotions.


Brian Tessellation came to Argentina to participate in the conference organized by the University of San Andres, the MIT Club and the Yale Club. Together with the Argentine Miguel San Martin, NASA researcher, they presented “Expansion of the Frontiers of Robotics and Automation”. TN Techno spoke to the Yale scientist about his work in the world of robotics.

– What is your current research about and the work you do at Yale?

– I am a computer scientist and I build robots. They are designed to tell us something about how humans and machines can interact with each other and sometimes that means learning about humans. Other times, it means we’re building technology so we can serve some valuable purpose.

Many of the robots I build are designed to teach or train or act like a therapist and are motivated by the idea that we don’t have trained teachers or therapists or trainers so that everyone can have one whenever they need, wherever they want, or demand it, in their homes.

Much of the work I did recently focused on building teaching tools for children. They can train 10-year-olds or five-year-old nutrition or assist six-month-old deaf children with sign language on math.

I am known for my work with children with autism,which is a social deficit. We build robots that come into houses and work with a child and his parents every day. They do the same kind of exercises and games that a therapist would do if people could pay for a specialist like that so they can sit with the kids and their parents every day.

Then for about 30 minutes a day the child will sit down and play with the robot and his parents. And they work with the same kind of therapy games they would do with a human therapist there. They are games that focus on anything from learning to take turns or learning to look people in the eye when you talk to them or recognize different emotions of someone’s face.

These robots express emotions. Especially when we’re teaching about emotions, the robot will model that behavior and say “that’s how I see myself when I’m sad” and it will look sad. And we’ll tell the boy to do the same. We have different robots, some more human and some more robots than. they look more like machines than people.

– How Wall-E?

– Sure, because even with Wall-e you can know when you are sad, because if it is so the head bends, the eyes narrow and the shoulders go down. You know if Wall-E is sad or happy. The same goes for our robots. We’re studying animator principles to make our robots move in a way that people will understand exactly what it means.

– Do you teach robots how to show emotions and then robots teach children how to do the same?

– Exactly. And in general, when we teach about emotions we will also show them images or videos on a screen, so that they can start to identify that not only with how the robot looks sad, but with what sad people look like. We are always very careful, because we don’t want to teach children just to respond to robots, we want to teach them how to respond to their parents.

That’s why every time we do these therapies/sessions, it’s the robot, the child and the parent because what the robot is really doing is linking the boy to his dad, so that then when the robot isn’t there, he still has these interactions with the parents.

– Because of these kinds of interactions, robots are sometimes treated like people? Kids sometimes do a voice search and ask for “please” and “thank you,” as if there’s someone inside, and the big ones just give the orders.

– From our point of view, we build robots that are designed to engage people and that’s why we want to treat the robot as if it were a social agent rather than an object. Whether it looks like a computer, toaster, or fridge, the important thing is to generate a desire to talk to you, that people are willing to attribute the same things to them as a person or a pet.

We found ways to encourage children and adults to treat this robot as if it were a social being. There are many tricks, some are based on the way they move, others are more subtle as a robot lying.

We make the robot play 20 minutes to “stone, paper or scissors”, 20 minutes is a long time, it is boring. But after 20 minutes suddenly the robot will only cheat on it once, shake your hand and, when you are losing, you will change it. You will see a stone and change the scissors for paper. And the robot says it won and people turn around immediately because they can’t believe it. This was just an object, immediately becomes “someone” and they start talking and make eye contact with him and shout and tell him “don’t do that again,” but they keep playing to see if the robot cheats again.


Leave a Reply

Your email address will not be published. Required fields are marked *