How AI Advances Are Improving Humanoid Robots

FLORA LICHTMAN: Hi, this is Flora Lichtman. And you’re listening to Science Friday.

[MUSIC PLAYING]

Today in the podcast, a conversation from our live show in Redwood City, California. Ira’s talking AI and what it means for humanoid robots.

KAREN LIU: The conversation part, it feels very amazing, but it’s actually the easiest part.

[APPLAUSE, CHEERING]

IRA FLATOW: Robots are just about everywhere these days. Maybe you’ve spotted one circling the grocery store, cleaning the floor at the airport, making deliveries, not to mention the robots on the assembly lines and factories. But what about having your own robot in your own home? How far are we from having a human-like robot at home? Joining me now to talk about the latest advances in humanoid robots is my guest, Dr. Karen Liu, professor of computer science at Stanford University, based, of course, in Stanford, California. Welcome to Science Friday.

KAREN LIU: Thank you, Ira.

IRA FLATOW: You’re welcome.

KAREN LIU: Nice to be here today.

[APPLAUSE]

IRA FLATOW: Now, you developed and trained humanoid robots. What is the definition of a humanoid? How do you define one?

KAREN LIU: Well, there’s really no universal consensus on the definition of a humanoid. But I think most people would agree that humanoids should have a similar morphology to humans. That means you should have two legs for locomotion, two hands for manipulation. And you need to be able to have egocentric vision systems so you can navigate the world. You could interact with objects in the world.

IRA FLATOW: Now, I understand not just talking about it. You’ve actually brought one.

KAREN LIU: Yeah.

IRA FLATOW: [INAUDIBLE], right?

KAREN LIU: We actually built a humanoid from scratch at Stanford.

IRA FLATOW: Toddy.

KAREN LIU: Toddy, as we call it, ToddlerBot, because it’s modeled off of a human toddler. It’s very small, but it has a lot of features, almost like a full-sized robot. It can talk. It can listen better than toddlers. And it can also see the world.

IRA FLATOW: All right, let’s bring him out. Let’s bring out Toddy, the toddler robot. Toddy, are you excited to be here tonight?

TODDY: Hehe. I’m so happy to be here too, buddy.

[LAUGHTER]

IRA FLATOW: Toddy, can you tell us how old you are?

TODDY: I’m 1 and 1/2 years old. Yuppy.

IRA FLATOW: Toddy, do you know that you are a robot?

TODDY: Hehe. Yep, I’m a robot toddler. Wiggle wiggly.

[LAUGHTER]

IRA FLATOW: I have to ask you something I ask all my two-year-old grandchildren. Toddy, do you have to go potty?

[LAUGHTER]

TODDY: Hehe. Nope, no potty for me. I’m all goody.

IRA FLATOW: Thank you for coming, Toddy. Good to see you. Thank you. Nice to see you, Toddy.

TODDY: Aw. Nice to see you, too, buddy.

IRA FLATOW: Thank you. Isn’t that great? That’s terrific.

[APPLAUSE]

KAREN LIU: [INAUDIBLE].

IRA FLATOW: Tell us, what are some of the advances in robotics that make something so humanoid like that?

KAREN LIU: Yeah. So there are so many things. AI is definitely one thing. I mean, you’ve probably heard of that. ChatGPT, large language models– they fundamentally changed the way robots think about the world, reason about the world. But there’s also something that is less mentioned, often overlooked factor, which is the advances in hardware, in robotics hardware.

And this is not so much about like one single breakthrough, technology breakthrough, but rather a convergence of all the enabling technologies. So for example, 3D printing. Like the ToddlerBot you just saw, they are completely made of 3D-printed materials. And all the motors are commercially available. So this is something that you can actually just build from scratch at home. And then I think that is one of the reasons that you would start to see that this is a very welcoming environment for robotics hobbyists to do this.

IRA FLATOW: How do you teach it to be intelligent like that?

KAREN LIU: Yeah, so that’s a really good question. The conversation part, it feels very amazing. But it’s actually the easiest part–

IRA FLATOW: Really?

KAREN LIU: –because of the ChatGPT, because of large language model. If we have an interface that can communicate with ChatGPT with a large model fast enough, then you can basically directly talk to ChatGPT. And so what you just did, we didn’t do any scripting, right?

IRA FLATOW: Sure.

KAREN LIU: I have no idea what–

IRA FLATOW: Right. You didn’t know what–

KAREN LIU: –Ira is going to ask.

IRA FLATOW: –I was going to ask of you.

KAREN LIU: I was a little bit worried there.

IRA FLATOW: [LAUGHS] I was more worried than you were, believe me.

KAREN LIU: Yeah, right? I was just– but I think the hardest part is something that you feel like it’s very easy to do, for example, balance, for example, walking forward, locomotion, manipulation. Those are things that we take for granted. But it’s very difficult to train robots to do that reliably.

IRA FLATOW: Is that the hardest part, then, to get it to walk and to mimic how humans do things?

KAREN LIU: Yeah. So walking is hard. Manipulation is another very difficult task. And the reason walking is really challenging for bipedal, meaning the robots with two legs, is because we are– this is what we call underactuated system. Underactuation means that we could actuate our joints, our motor– well, we don’t have motors, but humanoids use motors. We could actuate the motors locally. But the whole system, a whole dynamic system, we cannot accelerate it from point A to point B without using external forces.

So the way we taught humanoids can walk is because it figures out how to exert torques at its hip joint and knee joints and ankle joint in a very coordinated way, so that you can push the ground, get exactly the forces you want from the ground, and move your– accelerate your center of mass forward to the right location that you want it to be. And that’s a lot of computation there.

IRA FLATOW: I’ll bet. What about– I imagine the hands are just as difficult, right?

KAREN LIU: Yeah, hands, that’s a different thing. And the reason manipulation is really hard is because we really want to be able to manipulate anything in the world in a very general way. If you just build robot hands, robot arms to do one thing in the factory, in the industrial setting, that’s actually not too hard. But the reason it is hard today is because manipulation means that you need to handle all different kinds of diversity in the world, and do so with the same brand, same policy. That’s what we call it.

IRA FLATOW: Well, what do you think the limitations are of what you can teach Toddy or robots to do?

KAREN LIU: One of the biggest limitation is that robot doesn’t– like, if you were going for the machine learning or AI approach, we need a lot of the data to teach robots, just like the way we teach large language models. So my colleagues would say we have this 100,000 years of data deficiency comparing to training large language models. What it means is that in order to train a LLM, or large language models, it takes the data that takes an individual, a person, to read the text training data 100,000 years to read.

IRA FLATOW: Wow.

KAREN LIU: And if you think about all– the largest data set we have in robotics, it’s about 10,000 hours.

IRA FLATOW: 10,000 hours.

KAREN LIU: Yeah. That’s the data– among the data we have today.

IRA FLATOW: Right.

KAREN LIU: So data is definitely really challenging for robotics. And another really difficult thing is just where learning– not just understanding text or images. We need to understand the mapping between what you see and what you do because this is the most crucial part of the robotics, a decision-maker.

IRA FLATOW: Now we have lots of people lining up. I’ll try to hit both sides of the room. Let me start on this side first. Yes.

AUDIENCE: So you’ve talked about data to teach the robot. I’m a science teacher who teaches life science, and I think a lot about how our senses give us feedback, which is a huge part of learning how to walk for instance. What kinds of feedback does the robot take?

KAREN LIU: Yeah. So using data to teach robots– we call that imitation learning or supervised learning– is only one way. Robot can also learn from trial and error, learn from its own experiences of interacting with the world. And this is what we call reinforcement learning. In that case, we will ask robots just to try things. Here’s your task. You try different actions.

And then we look at the results. We’ll give you a score to let you know you’re doing well, or you’re not doing well. And robot will take that as a signal to decide, oh, what I just did was pretty bad. I got a poor score. Maybe I should reduce the probability of doing the same thing again. So that’s kind of like getting feedback from human.

But in order to make this process automatic, we don’t have a human keep telling robots what score you get. Instead, we would design what we call a reward function. It’s an automatic way to assess the performance of a robot.

IRA FLATOW: Next question, over here on this side.

AUDIENCE: So to follow up on the concept of reinforcement learning, I read, as a young boy, Isaac Asimov’s books. And among them were rules of robots. And so how does this reinforcement learning coordinate with that and the ethical use of robots to be sure they’re not going to harm us?

KAREN LIU: Yeah, so talking about ethical aspects of robotics, it is still– so the reward function I was just talking about is designed by a person. So if there are certain values and beliefs that we want the robot to learn from, we have to encode it into the reward function. And that is definitely not an easy way to do because how do you turn ethics into a mathematical function and equation?

IRA FLATOW: Do you think robots are going to get smarter than us and we’ll be the robots someday?

KAREN LIU: [CHUCKLES] Well, robots can appear very smart in certain ways, but sometimes they are also not so smart. So for example, going back to that reward function again, my colleague would say, if you asked a robot to cook dinner for you and you write it down into the reward function, robot would probably do that, cook dinner for you, but along the way, it kills your cat.

And the robot say, well, in the reward function, you say I get a good reward for cooking the dinner, and I don’t get any punishment for killing the cat. Why not? So the problem is that you can never– it’s kind of like arguing with a toddler. [LAUGHS] You didn’t say I cannot do that, right?

IRA FLATOW: Right.

KAREN LIU: There are always the corner cases that you cannot capture in your reward function.

IRA FLATOW: You don’t want to kill any cats, no. What happens when a robot fails? It can fail in many different ways. Can it actually teach itself what it did wrong without you having to tell it? Does it learn from its mistakes?

KAREN LIU: Yeah. That is actually a really great question, Ira, because I would say this is the major deficiency of the robot today, comparing to humans. And humans are really, really good at knowing how well– self-awareness, like knowing how well it does in this particular task, doing the skill learning time and after skill is deployed to the real world. If you ask a small child to put together a block structure and if the child fails at the task, it immediately knows what went wrong. It knows, oh, maybe I didn’t set a foundation sturdy enough, or I didn’t align this particular piece precise enough. Or a dog just notched the table and caused the structure to collapse. They know immediately.

And this ability, this innate ability of understanding the source of error is something that a robot doesn’t have. So robot will fail. And then the only thing it would know, it has no idea why it fails. It just, well, assume, I’m not going to do that same thing again because I got a bad score. But if robots can reflect on its behavior, maybe it will figure out, oh, because my sensor was not calibrated right, or I estimated the floor not as slippery as it should be. And then knowing that will help it to accelerate learning a lot.

IRA FLATOW: It’s not going to say, oh, darn, I did something that–

KAREN LIU: That’s right. Yeah.

IRA FLATOW: It’s not going to go that way.

[MUSIC PLAYING]

FLORA LICHTMAN: After the break, more of Ira’s conversation with roboticist Karen Liu about the current state of humanoid robots.

KAREN LIU: Yeah, so if you want something like Toddy, you can have it today.

IRA FLATOW: You can?

KAREN LIU: Oh, yeah.

[MUSIC PLAYING]

IRA FLATOW: You have a background in computer animation correctly.

KAREN LIU: Yes.

IRA FLATOW: How has that helped you in your robotics? What did you bring to that?

KAREN LIU: That’s right. So I actually started as a computer graphics person and then worked my way into robotics. But if you think about computer animation, you probably imagine– or there is animators making those keyframe, those poses, and then they interpolate those keyframes and make a nice-looking, smooth animation. And that is something that you can do for, let’s say, making movies. But if you want to do video games, then you probably will need a process and automatic process that can generate motion– let’s just call that motion generator– which understand the high level commands like moving forward, make it jump on the platform when you play video games.

And if you take one step further, instead of this motion generator, instead of outputting a pose or a sequence of poses, it outputs joint torques. Then what you can do is you take that joint torque, together with the gravity and contact forces, whatever external forces is, put it into the physics engine, and let the physics engine figure out the poses. And this is what we call physics-based animation.

And at this point, you are actually very close to robotics. The only difference is that you have human animators that given those high-level commands, or you have an AI model to do that. So if you replace the humans with an AI model and let this AI model make high-level commands based on the task, based on the observation, then this is basically– your character is basically a virtual robot. And you are ready to deploy it to the real world.

IRA FLATOW: Now, I understand that Toddy, who was out before, was actually a 3D-printed–

KAREN LIU: Yeah.

IRA FLATOW: –body?

KAREN LIU: Yes.

IRA FLATOW: Which makes me think, how far along until we can actually get our own home robot, do you think?

KAREN LIU: [CHUCKLES] Yeah. So if you want something like Toddy, you can have it today.

IRA FLATOW: You can?

KAREN LIU: Oh, yeah. So–

IRA FLATOW: Do we have them in the lobby for sale out there? No.

KAREN LIU: No, but almost as good as that. We open source everything. We open source the hardware design, the algorithms, and also step-by-step instruction manual. So if you have a 3D printer at home, you want to build Toddy, all you need to do is just order those commercially available motors and then just follow the instructions. And so–

IRA FLATOW: Sure. If you can bake a cake, you can do that, right?

[LAUGHTER]

KAREN LIU: Yeah, we have the recipe.

IRA FLATOW: [LAUGHS]

KAREN LIU: Yeah.

IRA FLATOW: But in general, when might we see home robots? Not maybe just Toddy, but other robots.

KAREN LIU: It is a really difficult question, and I get this question all the time. And I always say five years. [LAUGHS]

IRA FLATOW: Everything’s always five years away, right?

KAREN LIU: That’s right. Because beyond five years, then it’s not my responsibility.

IRA FLATOW: [LAUGHS]

KAREN LIU: The reason I think it is possible, but not just tomorrow is because right now, we’re still building a lot of the building blocks, like locomotion, manipulation, before they are completely reliable, robust, and safe. I don’t think it is ready to build a product out of it.

But I also feel like the deployment of humanoids is going to come in stages, kind of like autonomous vehicles. So at the very beginning, maybe the robots are not going to be able to do anything you want in your household, maybe just for laundry or clean your dish– dishwasher.

IRA FLATOW: I’d settle for that.

KAREN LIU: I know.

IRA FLATOW: Loading the dishwasher. Yeah.

KAREN LIU: Those are the two things I want.

IRA FLATOW: Dr. Liu, this is fascinating. Thank you for taking time to be with us today.

KAREN LIU: Oh, of course.

IRA FLATOW: Dr. Karen Liu, professor of computer science at Stanford–

KAREN LIU: Thank you so much.

IRA FLATOW: –based, of course, in Stanford, California.

[CHEERING, APPLAUSE]

[MUSIC PLAYING]

FLORA LICHTMAN: Thanks for listening. Don’t forget to rate and review us wherever you listen. It really does help us get the word out and get the show in front of new listeners. Today’s episode was produced by Shoshannah Buxbaum. I’m Flora Lichtman. Thanks for listening.

Copyright © 2025 Science Friday Initiative. All rights reserved. Science Friday transcripts are produced on a tight deadline by 3Play Media. Fidelity to the original aired/published audio or video file might vary, and text might be updated or amended in the future. For the authoritative record of Science Friday’s programming, please visit the original aired/published recording. For terms of use and more information, visit our policies pages at http://www.sciencefriday.com/about/policies/