Baymax, the Man Who Invented Him, and Other Stories From the World of Robotics

The humor and wisdom of a scholar.

You may already know Professor Chris Atkeson. He has taught at the Robotics Institute of Carnegie Mellon University since 2000, and before that (some 20-odd years ago) he taught at the Massachusetts Institute of Technology. This morning QDaily published an article he wrote about the robot that inspired Baymax ("Dabai" to Chinese audiences) in Big Hero 6: what it really looks like, why he designed a soft robot like it, and how close a real robot could come to Baymax's intelligence in the foreseeable future.

We enjoyed talking to Chris, and one big reason is that he does not see himself as the fabled "father of Baymax." The label may come from a profile published by 36kr. As the Q&A below shows, neither the "collaboration" with Disney nor the public's passion for Baymax has changed him much.

He is still raising money to build a real Baymax: a soft robot that cares for people's health.

The Robotics Institute at Carnegie Mellon, where Atkeson works, is the largest in the world, and Carnegie Mellon is one of the few universities with a stand-alone robotics institute. It has 76 faculty members, 94 PhD students and 132 master's students. Its alumni are as impressive as the institute itself. They include Chris Urmson, who started Google's self-driving car project; Boris Sofman, Mark Palatucci and Hanns Tappeiner, co-founders of Anki, a smart-car development platform; and Mark Maimone of NASA's Curiosity Mars rover team.

QDaily ("Q") interviewed Professor Chris Atkeson ("A"). What follows are more stories about Baymax, about Atkeson himself, about soft robots and about robotics research.

Professor Chris Atkeson and Baymax

Q: Did you know you now have quite a few fans in China? Some articles even call you the "father of Baymax." How do you feel about that?

A: The real fathers of Baymax are the Disney filmmakers. They got a bag of ideas from me and created a whole lot of characters out of it. All I did was hand them the bag.

Q: I'd love to know what went through your mind the first time you saw a robot from your own field up on the big screen, looking so futuristic.

A: It's beautiful. It isn't real, but it was done very well, and it showed me what the real thing (a soft robot) might look like one day.

Q: How does the Baymax in the film differ from the robots you imagine serving people in the future?

A: Disney decided Baymax had to hold up a huge rock and so had to be very strong, so they gave him a skeleton. The robot we are designing is meant to be very light, so it has no skeleton.

Q: So your robot has no bones? How does it move, then?

A: It has electric motors that work like muscles. Many animals move by muscle (think of an elephant's trunk or an earthworm). Some of their soft tissue also gives support.

Q: If you were the director or the character designer, would you change anything else about Baymax?

A: I'd change Baymax's legs. It would be better if he had knees.

Q: Do you think Baymax is too big?

A: I honestly don't know why they made Baymax so big. There may be artistic reasons, but I think you may be right that he could be a bit smaller.

Q: It seems it all began in 2011, when Don Hall, one of the co-directors of Big Hero 6, visited your lab. The next thing we knew, Big Hero 6 was out and everyone went to see it... No one seems to know what happened between you in the years in between. Can you tell us?

A: In fact we had very little contact in those years. After they got the idea for Baymax, they went off and made the (film), and they did it very well.

Q: Did they come back to ask you about soft robotics?

A: No. On the whole, it was their idea, and it's their robot.

Q: I see. That's not quite what we imagined. Plenty of Hollywood films have featured robots, from R2-D2 in Star Wars to WALL-E, but none seems to have drawn as much attention and affection as Baymax. What's different this time?

A: I'm not sure I agree. I think R2-D2 and some other robots are very famous too; there are plenty of very famous robots. As I see it, Baymax is a friendly, adorable robot, and lots of people like him. One interesting thing is that in many countries, in Europe for example, people fear technology and robots. In the United States we're fairly matter-of-fact about them, and in Japan people love robots. I'm not sure whether people in China like robots.

Q: Hmm, it's a bit complicated. Before Baymax, the robots we saw were all metal and big, and some had sharp edges...

A: Right. I usually call robots like that "Metal Monsters."

Q: In your essay "Why I Want To Build Personal Health Care Companions Like Baymax," you wrote that Baymax raised people's hopes, as if such a robot might really appear one day. How far has research on soft robots come, especially in medicine?

A: Not much work has been done on soft robots overall. The field is just getting started. People already build metal robots fairly well, but soft robots aren't there yet. We need to make soft robots look strong and dexterous too.

Q: Why do you think metal robots came before soft ones? Or, put another way, why is soft-robot research harder?

A: Soft robotics is a new idea and a new way of making things (of thinking and solving problems). For engineers, changing the way they work is hard. If you build a metal robot, the joint design is easy to understand, but building a soft robot is more complicated.

Q: Do you know of any labs in Japan doing similar research?

A: I know there's some work on soft robots in Japan, but I don't know exactly who is doing it. I know of some in the United States. A lab in San Francisco called Other Lab is doing very well.

Q: Why are people keen to build soft robots? What's the biggest difference from metal robots?

A: Safety. If you want a robot to come close to people, or touch them, a metal robot is not very safe. There's another important reason, and it may favor China: the way we build soft robots is very similar to the way clothing and fabrics are made. So, compared with a car, a soft robot's price may be closer to that of clothing. Chinese makers of clothing and plush toys are very capable, so we believe China could become a leader in soft robotics.

Q: That's good news. Are you working with other professors or companies in this field?

A: Yes. We work with a US company called iRobot, some other labs in San Francisco and some other CMU professors who study soft robots.

Q: Those are all robotics people. Is anyone from clothing and fabrics involved?

A: Not yet. Some other labs pay more attention to craft and fabrics, but we haven't put much emphasis on that so far.

Q: Back to Baymax. The Baymax in the film is very intelligent and good at talking to people. What still separates today's real soft robots from Baymax?

A: Artificial intelligence, the robot's brain, is the bigger problem. One interesting question is where the robot's brain will come from. I think it will most likely come from something like the voice assistants on phones: Siri, or the assistant Amazon makes, and there's also software on Android you can talk to. A great many people are working to make phones smarter, and in the end that may make robots smarter too.

Q: So it's like Baymax with an iPhone built in?

A: Right. It's really the same computer parts. You can make them into a phone, a voice assistant or a robot.

Q: Since Don Hall's visit in 2011, has your field made any progress? Have any problems that were open then been solved?

A: Back then we didn't know whether we could build a strong soft robot. Working with other labs, we built one with real strength, and we're very happy about that. Then we started work on the robot's skin, to give it a sense of touch. The film never shows Baymax sensing things through his skin, but for a robot this is very important.

Q: We know your research inspired Disney. What inspired you and your research?

A: Hmm... when I think about humans and robots, I think of sailboats. There are soft sails and ropes, and there are spars too. As for the body, it's like what you see when you eat meat: there's muscle, soft tissue, bone... I wanted to build a robot like a sailboat.

Q: We've learned a lot about Baymax. Now let's talk about the "father of Baymax." Wikipedia doesn't say much about you, so I'm curious: when did your interest in robots begin?

A: I come from a big family, with seven brothers and one sister, and all my brothers were athletes. But I wasn't much good at sports. I was slow, a bit heavy and not very agile. So I got interested in making machines that could do what my brothers did.

Q: Like the robot fights at the start of the film?

A: Not exactly. There are fighting robots, but I was more interested in the Olympics: getting robots to run, swim and so on.

Q: When did you build your first robot?

A: 1981. But keep in mind that I'm very old, and before 1980 there wasn't much you could do with robots. It was very hard for a kid to build a robot back then. Now it's become easier.

Q: Well, in 1980 that made you a pioneer.

A: Yes, that was a long time ago. There were hardly any computers or robots then.

Q: Many people loved robots when they were young, through films or novels, but ended up as lawyers, doctors and so on. How did you get into this line of work and make it your life's career?

A: I was lucky. In college I wanted to know how the brain works, and I was lucky enough to go to MIT for graduate (PhD) study. Along the way I found that one way to understand how the human brain works is to build robots. So I started doing research in robotics.

Q: So you moved into computer science?

A: Into the Robotics Institute, to be precise. It sits within the broader field of computer science, but at the Robotics Institute we focus mainly on robots.

Q: Do you think robotics research requires an understanding of the brain?

A: I don't think it's strictly necessary, but it's very important. I worry about the many elderly people who need help. If we knew why old people fall, maybe we could help them fall less. So if you want to understand robots, I think it's important to understand how humans work.

Q: So was it your concern for the elderly that turned your research toward soft robots? Because they're friendlier and safer, which suits slow-moving older people?

A: Let me put it this way. My grandfather had an illness, and he sometimes fell, and my grandmother wasn't strong enough to pick him up. If a robot had been there, it would have helped a great deal. My grandmother could have told the robot what to do, and it could have caught my grandfather before he fell.

Q: That matters. I also know that many scientists find their research questions in their own lives and the lives of those closest to them. Your page shows that you went from MIT to Georgia Tech and then settled at CMU, where you have now been for 14 years. Some say MIT focuses more on specific fields while CMU has more diversity. Do you agree?

A: Simply put, I think one important reason is that CMU has a great many people working on robotics. MIT and Stanford have some too, but they're scattered across different departments, so perhaps they can't talk to each other often.

Q: As far as I know, CMU's Robotics Institute is the largest robotics institute. Is that right?

A: I think so.

Q: It sounds like I should pay a visit.

A: There are lots of Chinese students here, and I have a lot of Chinese students too. Most of them have engineering backgrounds.

Q: So was this community also a reason you came to CMU?

A: Yes.

Q: Can you tell us about your lab? Is it much like the one in the film?

A: The Disney team visited many labs and combined elements of them to create the brand-new lab in the film. Our lab isn't exactly like the one in the film, but it's very similar.

Q: So it's also a big open space, with lots of freedom and energy? Do you know where else they visited?

A: Yes. They visited MIT, and Harvard, I'm not quite sure, and many schools in Japan, such as the University of Tokyo, and some others.

Q: And they blended them all into one very fancy lab?

A: Yes. I think they ended up making the lab of their dreams.

Q: A CMU student told us you're a very efficient teacher but also a very strict one. Do you agree?

A: Strict? Well, okay, you could say that... I am fairly strict.

Q: What qualities do you look for when choosing PhD students?

A: They need to be very curious and passionate about what they do.

Q: And demanding of themselves in their work?

A: Yes.

Q: How do you choose students? Your homepage says not to apply to you directly by email, and that applicants must first get through the university's selection. Among those who have, how do you choose?

A: I talk with them, listen to what they say and look at what they've built. If we can find a project to work on together, I'll consider them.

Q: How many PhD students do you have?

A: I had six students last year. Two graduated this year, so four remain.

Q: Do you work together often?

A: Oh, I try to give them a lot of freedom, but we also try to meet and talk often.

Q: Are they all working on soft robots?

A: No. The ones working on soft robots have graduated. The rest are working on Metal Monsters, for something called The Robotics Challenge. It will be on show in June.

Q: I saw that the soft-robot project also belongs to a CMU center called the Quality of Life Technology Center. Can you tell us more about it?

A: That's right. The Center's goal is to use technology to help people live better, for example by helping disabled and elderly people live in their own homes longer (instead of being sent to hospitals and nursing homes). I'm one of the professors there working on robots, and many others work on other areas.

Q: How many people are in the Center?

A: I'm not sure. About 50? A lot of people.

Q: Are they all CMU professors?

A: Yes, and some professors from the University of Pittsburgh as well.

Q: Ah, I know the University of Pittsburgh has a very good medical school.

A: That's right. For us, building ties with medicine is very important too.

Q: Do you think Baymax will finally emerge from a lab, or perhaps from a factory?

A: That's a very interesting question, and I don't know the answer. In the past, robots were mostly designed in labs, but things have become more complicated. In the United States, some companies, such as Google, have bought robotics companies. Recently a company bought part of CMU, too. So perhaps more of the work will be done inside companies in the future, and we'll end up like today's aerospace industry: aircraft companies build aircraft, while universities research particular problems or train experts but don't build aircraft themselves.

Q: Right. I know places like Microsoft Research are doing similar things.

A: Yes, that's possible too.

Q: We know your two research areas are humanoid robotics and human-aware environments, and Baymax combines the two well. This field is very specialized for me, so I can only make guesses. For example, it seems to me the second is harder to achieve: first the machine has to understand human speech, and then it has to learn on its own from its surroundings, and so on. Is that right?

A: I think the environment may arrive before the robot. For example, you set up your iPhone at home to record video. It can watch you and try to understand what's going on.

Q: And then learn from its surroundings?

A: Yes.

Q: So you think the environment will come before the robot?

A: Yes. In the United States some companies already make home monitoring systems to keep elderly people from getting into trouble at home.

Q: Sounds like an intelligent environment?

A: Exactly. That's it.

Q: We mentioned Siri, which is sometimes not that smart. So are we still a long way from building a real Baymax?

A: Yes. Maybe we'll first have a fairly dumb robot, and it will eventually become smart.

Q: And we'll watch it get a little smarter at a time, until it becomes Baymax?

A: We can accept a dumber robot because it is very soft and safe. A (metal robot) that dumb would be rather dangerous.

Q: We saw the Build Baymax page. It looks like a crowdfunding campaign.

A: I hope it can become one. It's also the first time I've drawn such a crowd, haha. People are very excited and say they want to do something to help make Baymax real.

Q: Has anyone donated?

A: Yes, some people already have.

Q: I'm guessing the page hasn't been up long.

A: Right, not long. Only two months.

Q: Getting attention that quickly looks like good news. How many people are working on the project now?

A: As I said, the people who worked on Baymax have graduated, so now I need new people, including from China. But he or she may need a visa first!

Q: Who funded the project before now?

A: The National Science Foundation. The grant ran for five years and has now ended, so I need new funding.

Q: This may be hard to answer, but how long do you think it will be before we have a real Baymax?

A: Do you mean the body or the brain?

Q: Both.

A: The body should come soon, within the next few years. As for the brain, as you said, Siri still isn't that smart, and I quite agree with you. So that may take a long time.

Q: But before then we could still have some not-so-smart robots helping elderly people up?

A: Uh-huh. Take a couple, for example: if one of them is about to fall, the robot may be fairly dumb, but the other person can tell it what to do.

Sun Chunyang also contributed to this article. Sun is a graduate student in applied robotics at Carnegie Mellon University and is helping the professor raise money in China for Build Baymax. To get in touch, write to [email protected] or search for the WeChat official account buildbaymax.