Soft Robots Navigating Mazes Without Human Guidance with Jie Yin, Associate Professor at North Carolina State University

(Intro Narrator at 00:00:03) Hello, my friends. Today we're talking to Ji Yin, associate professor at North Carolina State University, and we discuss Ji's autonomous noodle-shaped robot, the blueprints of soft robotics, and the impact that this research could have on the world. All of this right here, right now on the Modern CTO podcast.

(Joel Beasley at 00:00:29) This is the Modern CTO podcast. I was checking out the video. It's like a piece of pasta that's twisting and moving and getting itself out of a maze. What is that?

(Ji Yin at 00:00:47) Okay, so this is we call it a new type of soft robot, but this software different from previous soft robots. This robot is autonomous. You see, it's going to self-roll without any other — without electricity, right? So if you put this guy on the hot surfaces, then it's going to self-roll. By "self-roll," right, I mean it's going to roll by itself. And then the interesting thing here is whenever it encounters obstacles, then it can change its moving directions. So it has two ways to change that. One is it's going to circle around the obstacles, and then the other one is going to be — we call it the snap-back. So it's going to revert its direction, the moving direction. So for example, if you move forward and then when it meets an obstacle, it's going to move backward autonomously without any external human intervention or computer controls.

(Joel Beasley at 00:01:47) So there's no electronics inside of it?

(Ji Yin at 00:01:48) Yeah, no electronics inside it. Definitely.

(Joel Beasley at 00:01:51) Then how is that possible? What is this substance?

(Ji Yin at 00:01:55) Yeah, this is a very good question. So you see, this soft robot is made of we call it the smart materials. So this material actually can respond to the heat, to the light. This material is called the LCE. It's a liquid crystal elastomer. So you see the computers while we're using the computer, the screen is a LCD, right? It's a liquid crystal display. But this material is kind of similar. It's called a liquid crystal elastomer. It's very soft. You know, it's sometimes even softer than our skin. So the good thing about this material, like what I just mentioned, right, so this material can respond to the heat. So for example, if we're shining a light on this one, on this piece of material, it's going to shrink. So rather than expand, actually, it's going to shrink. Or if you put this guy on the hot plate, it's going to start to shrink. So the size, depending on how high the temperature, it can shrink by half sometimes. So now if you make this guy, right, so become a twist, like what I just mentioned, the pasta. So it's like a rotini. You know, you make a twist. And then actually after you make a twist, and then you can lock the shape into a twist shape. You know, in room temperature, it's like a pasta, right? So it's a twist shape. And now if you put it on the hot surfaces, it's going to start to untwist itself a little bit, and also it's going to self-roll like what I just mentioned. You know, so the shape will not be a straight one. Now it's a little bit slightly bended. You know, it's like an arc shape. It's going to bend that forward and then keep moving.

(Joel Beasley at 00:03:29) Does it have awareness in the sense that you show it navigating a maze and it gets out? If you give it a different-shaped maze, would it know which of the two mazes it's in?

(Ji Yin at 00:03:43) Yeah, that's a good question. So actually we have tried different types of mazes in terms of, you know, how complex. For example, the complex means we can add more channels, right, into the maze. I would say this is we call it a simple maze, but it's not a really — it's definitely not like a real maze. So we compare two different mazes, and then we find that, you know, depends on which location you put it on, right? I mean, the escaping time can be very different. You know, sometime take ten minutes, sometime take two minutes, and you know, sometimes it can take half an hour or two to get it out, to find a way out. But, you know, here's what I think. You know, for this one, right, so if this guy moves toward a wall, right, if its motion is blocked by a wall, then it's going to snap back, you know, and then change direction. And then sometimes it could be bounced back and forth, but you know, and then it can change its directions. Okay, so always it can find its way out as long as there's an exit. So this is more like an iRobot. So, the iRobot, right? So you have everything there, and then you — if you encounter a wall, right, you're going to have to change. You know, you bounce back and then you go all the way.

(Joel Beasley at 00:04:58) That is pretty cool. And so have you ever tried putting it in a circle, a maze that it can't get out of?

(Ji Yin at 00:05:03) We haven't tried the more complex one, but I mean, unfortunately, actually it cannot escape because, you know, sometimes, like what I just mentioned, right, sometimes if you have something like a panel, you know, the walls, right, and then this guy just bounces back and forth between these two panel walls, and then it cannot get out. So that's actually one of our future directions we're working on. We want to build something like a self-turning capability because, you know, for this one, you cannot make a turn if without any obstacles, right? So that's, you know, whenever you make a change, you just rely on the obstacle, the interaction between the soft robot and the walls.

(Joel Beasley at 00:05:47) And what makes it a robot? Like, why that word?

(Ji Yin at 00:05:51) Yeah, I know. The first, right? The first look at this noodle-like robot. Okay, is this really a robot? Yeah, it depends on, right, so how you define the robot. So to me, a robot is, for example, a soft robot. Right, so first it's made up of soft materials, right? It's not a hard material. And then the second thing is, you know, it can do the actuation. You know, it can sense the environment. Action means the actuation. So actuation means it can deform and it can bend. You know, it can change its body shape, and then it can generate the motions, right? So sensing, actuation, and motion. And then we add one more thing. This is called decision-making because if you put it in a confined space, this robot is trying to find its way out. So this is called a very low level of intelligence. So we call this self-decision-making because it does not require any external control or any human intervention inside. I think, you know, the definition of robot is going to, nowadays, right, become very broad.

(Joel Beasley at 00:07:00) Oh yeah, I agree. I mean, I get to research all this amazing technology, and it's always interesting because when you see it, I agree with soft robot. And then I ask myself, well, then what's a robot? And like, how do you define robot? But I'm cool with the word, just so you know. I'm on your team. I think it's a soft robot. Why is it driven to get out? Like, why isn't it driven to dance?

(Ji Yin at 00:07:22) Well, you're asking excellent questions. So currently we're working on one piece of paper. It's called a dancing robot. So it's really — it's like, you know, you can flick. You know, you can change its shape, how long it can dance, but rather than go out of a maze, you know, like that.

(Joel Beasley at 00:07:42) Okay, but you don't know, like, the fundamental reason of, like, why the noodle robot wants to escape. You just put the materials together and then that's what it did. That's just how it operates.

(Ji Yin at 00:07:53) Actually, it's a — yeah, I'm a — so it's a long story. So in the beginning, right, actually we find that this kind of snapping is kind of surprising because, you know, you know that snapping is more like a Venus flytrap, right? If you know the Venus flytrap. So the fly comes in, it's going to snap, right, to close its leaves, right, to catch a bug like that, right? So this — I mean, the snapping process is very short. It's only twenty to two hundred milliseconds. So this one is the same. So we have this twist one, right? We put it on a hot plate. We find, oh, this guy can self-roll. You know, it can self-roll. The reason is because there's a temperature gradient because on the bottom it's going to be heated, right? So that means the bottom part is going to shrink, but on the top it's not in contact with any, you know, the heat source, right, the hot surface. And then you generate a temperature gradient from bottom — it's high, top is low. In this case, you're going to drive a deformation, so you can see it's going to bend like, you know, like an arc, right? Bend like that, and then start to roll. See, it starts to roll because the top one becomes hot. You know, the top one is cool. The cool on the top becomes hot. You know, the bottom one becomes cool. Now you can self-roll without stopping.

(Joel Beasley at 00:09:11) If you and I were to go into a lab and make one of these things, how would we do it?

(Ji Yin at 00:09:19) Oh, okay. We'll make these materials, right? So we should take two steps. So step one is, you know, you make your synthesized materials, right? And then you just cure it — like a gum-like material, right? Now you start to stretch. Okay, you stretch it. And then now you twist it and hold it. Okay, stretch and twist. Hold it. Because now it'll give you the twisted shape, right? Okay, now hold it, and then now you're going to do a second step of curing. So that means you're going to have this guy exposed to UV light. So the UV light can help to cure the material to make this material become solid. And then after curing for some time, and then it's done. I mean, the shape is fixed. Now it's become a twist.

(Joel Beasley at 00:10:06) That's pretty cool. And then how does this connect to, like, commercial application? Can you have buildings that assemble themselves, or, like, what can you do with this way off in the future?

(Ji Yin at 00:10:18) Yeah, that's actually — I mean, we're doing the research, so we didn't think too much about the commercial application like that. But scientifically, I think there will be two — I think it's a twofold. So one is, you know, if we can send this kind of soft, autonomous, and intelligent robot to a harsh environment, right? Harsh environment means an environment that has extremely high temperature, for example, like a desert, right? So, you know, the global warming — with the global warming, the desert, the temperature, the surface, they call it the land surface temperature, it's going to change from 52 to 52 degrees C to 81 degrees C. That's the hottest one. So that means, you know, in that way, people definitely will not send the people there, right? So you want the robots to do some health — for example, like environmental monitoring, right, or navigation, like that. So now you want to find this kind of soft robot, right? So you see we have a slope. So this soft robot, you know, you can go there, and then you can integrate some — we call it a wireless sensor, right? Wireless sensor, you know, on that, and it can send some signal back. So in this case, they can take the firsthand information — for example, the surface temperature or maybe, you know, some other, for example, terrains, whatever, you know, those kind of different things. You know, that's one of the applications.

(Joel Beasley at 00:11:53) And so you said what's driving the motion is the temperature gradient. So it's going towards the warmth? Yeah. Okay, so it's survival. That's what it's going for.

(Ji Yin at 00:12:03) Yes, yes, exactly.

(Joel Beasley at 00:12:04) It's cool how that's baked down even in the lowest level of our things that exhibit some sort of intelligence.

(Ji Yin at 00:12:11) Yes, yes. You know, I forgot to mention another point, you know, for the potential application. I think this is very exciting, actually. So I was thinking about the outer space, right? For example, the moon. If you check the temperature on the moon, you know, the daytime temperature, it can go like 120 degrees C. So for our soft robot, the working temperature — the working temperature means, you know, on this range, the soft robot is going to self-roll, right? So working temperature is between 52 to 180 degrees C. So you see, on the moon, right, so the highest temperature around like 120 degrees C actually is within the working temperature of our soft robot. So I was thinking, right, so if you scroll this guy to the, you know, to the moon on the daytime, actually it's going to — yeah, it's going to watch it to just self-roll and then to go all around.

(Joel Beasley at 00:13:04) That's pretty cool. Yeah. Are all soft robots made out of the same base material?

(Ji Yin at 00:13:09) So this, you know, for our soft robot, yes, it's made of these materials. But, you know, for other researchers, right, so they are trying to explore a new kind of materials, soft materials. We call it soft, active materials. For example, you know, the hydrogels or magnetic elastomers. For example, you have ferro particles, right? You put some ferro particles inside of the elastomers, right? And now, you know, it becomes — you can do magnetization. Now it's become you can use magnetic field, right, to drive the motions like that. So these are all different types of soft materials that are widely, you know, used in the soft robotics community.

(Joel Beasley at 00:13:53) That is so cool. And then you work on a couple projects, something that might have a potential for water collection. Can you explain what that is?

(Ji Yin at 00:14:01) Oh, okay. So yeah, the water collection is actually it's for the harvesting. So that's related to the surface, you know, the wetting behavior of the surface, right? So for example, the lotus leaf, right? You know, the lotus leaf is like a repair waters, right? So this is a kind of harvesting the water vapors, you know, in the air and then turning it into water.

(Joel Beasley at 00:14:27) And then are any of these robots programmable in the traditional sense? Like a robot that might be able to do three or four things, but I tell it to do this one specific thing.

(Ji Yin at 00:14:40) Oh, okay. So I mentioned it's kind of multitasking. For this one, I mean, so far we actually do not integrate any functionality into this soft robot. So the functionality means we do not integrate any sensors, right? So those kind of flexible sensors, you know, those kind of small-size sensors on it. For example, you can have a temperature sensor. You can have a moisture sensor. You can have, like, you know, those kind of gas sensors, right? There's a lot of types of sensors. It depends on what kind of reaction you want to do. So my point is, you know, this is more like a vehicle, right? So it's kind of an autonomous, intelligent vehicle. On this vehicle, you can have a lot of passengers, right? For example, all these sensors, like a passenger — you can have different sensors put on, you know, integrated together. I'm definitely should've been in a wireless way. And then they can send a signal back like that. Yeah, and then after that, it's going to maybe become multitask and also multifunctional.

(Joel Beasley at 00:15:51) Is there smart materials that move in ways other than towards the warmth of a temperature gradient?

(Ji Yin at 00:16:00) No, this is a good question. You know, these ideas may work for other soft materials as long as, you know, it can respond to heat, right? As long as it can either shrink or either expand to some extent.

(Joel Beasley at 00:16:18) I want to share my screen if that's okay.

(Ji Yin at 00:16:20) Yeah, sure.

(Joel Beasley at 00:16:21) Okay, this is — I don't know how to say the word. It starts with a K.

(Ji Yin at 00:16:25) But it's a kirigami.

(Joel Beasley at 00:16:27) Kirigami, okay.

(Ji Yin at 00:16:29) It's like, you know, origami. Origami, you know, origami, right?

(Joel Beasley at 00:16:33) Yes. Yes.

(Ji Yin at 00:16:33) So kirigami is, you know, ori means the fold. Kiri means the cut.

(Joel Beasley at 00:16:40) I like that. Very cool. So what is this thing doing?

(Ji Yin at 00:16:44) So this is a piece of, I mean, a lot of piece of our work. So we call it long, destructive grippers. But you can think, all right, if now if you are trying to grab an egg, the grippers, you know, you want to grasp an egg. That's how the soft grippers can do. Right? Because soft gripper, you can apply a small force and then we don't damage that one. Right? So actually, for this work, we were trying to challenge ourselves, say, how about I remove the shell? I just, you know, just we call it a raw egg. So the raw egg is like a fluid. Right? It's even softer than tofu. You know? How can you grasp a raw egg without damaging it? So that's the motivation.

(Ji Yin at 00:17:32) So now we're trying to think, well, okay. Definitely, we cannot pinch it. So we are using a way just like both hands that you encapsulate. You know, it's like encapsulate that and then to go from the bottom to pick up. So this gripper is made of, I mean, you can make them from paper or made of other, you know, plastic sheets. And then now I can start to grasp a raw egg, yeah, without damage because it's not pinching the egg.

(Joel Beasley at 00:18:00) Yeah. When I see all the different videos on your page, this one stands out to me as the one that looks like it has the most commercial potential because it seems like we would have a need for doing that. But of all of these things, which do you think will ultimately change the world the most?

(Ji Yin at 00:18:19) Well, these are very big questions. So I think, you know, the kirigami gripper, it may change the way we manipulate those kind of extreme, fragile objects, you know, fully delicate objects.

(Joel Beasley at 00:18:39) Yeah. That looks really cool. I think that one will be really big. And then how do you get paid to do this? Like, how do you get paid to do this research and hang out and find out these cool discoveries?

(Ji Yin at 00:18:51) Yeah. Definitely, I need to thank the funding support from the NSF, the National Science Foundation, because, you know, we have found some interesting studies. Right? And then we're going to apply for this grant and to support our research.

(Joel Beasley at 00:19:06) Very cool. Very cool. And then you did your postdoc at MIT. What did you study there? Soft robots?

(Ji Yin at 00:19:12) No. Actually, no. So I got my PhD in 2010 at Columbia University. So it's doing some mechanics things. And after that, like what I mentioned, right, I said, one time I did my postdoc there. So it has nothing to do with soft robots. And then I began my independent career, right, as a faculty, as a researcher at the Temple University in Philadelphia, since 2013. Because now I'm independent. Right? I have my own group. I have my own students. I'm trying to do some exciting things. And then I found that, and I found it, soft robots actually is a very interesting topic because this topic is new. It's not like a rigid robot has been worked on, like, say, like fifty or a hundred years. Right? But soft robots is has a very, it's a very young age. So it's about, I mean, the history of soft robots maybe only, like, twenty years. So that means, you know, the new things will have a lot of potential opportunities to, you know, to grow and to make it become more impactful to the society. Because, you know, if you see, for the commercial products, right, so if you see them on the market, it's not that many, you know, products, right, related to soft robots. But you see definitely there's a lot of rigid robots, all this kind of stuff. Right? But for the soft robots, really, we didn't see too many commercial products. So maybe that's the way, you know, the soft robotics community will also want to, you know, to work hard and work hard, you know, to commercialize, you know, some things because I think a soft robot is more like a complimentary, you know, to the rigid robots.

(Joel Beasley at 00:20:54) That is so cool. I'm so glad that people like you exist because, you know, you're very smart. You're studying these things. I've gotten to talk with people, and some of our greatest discoveries came from people just exploring and searching. And then, you know, you get the researchers and the scientists, and then you have the business people that can come along, and we all sort of work together. And so I'm kind of split. I'm kind of like fifty-fifty, but I really enjoy when I get to meet people like you who are really brilliant at what you do and can help explain these things to me because I'm a curious mind, and a lot of other people are too. And I'm excited to see what happens with soft robots. And as it progresses, let us know. Like, as new things come about, like, reach out to us and say, hey. Some new stuff happened with the soft robots, and they're conscious now and they're taking over the world, so lock your door.

(Ji Yin at 00:21:48) Yes. Yes. Yeah. Definitely. Definitely. Yes. Also, yeah, I want to thank you guys, you know, for to bridge the science or the scientist with the audience. Right? Because maybe the audience, they are not all of them are doing research, but, you know, they feel this is fun. I think, you know, if they feel this is fun, I know, okay. Maybe some of them, they think maybe in the future, they want to also do some scientific research or even, you know, in this soft robotics area, that'd be great because we do need new blood, you know, the new generations or future generations, you know, to get involved in this field and then to move forward and then to make a really huge impact to the society like that.

(Joel Beasley at 00:22:31) Yeah. That's cool. Someone out there is going to find some amazing way to change the world with this technology.

(Ji Yin at 00:22:36) Yeah. Yes. Yes.

(Joel Beasley at 00:22:38) Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you'd like to hear discussed on the podcast, either add me on LinkedIn or send me an email [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.