The Real Reason Robots Shouldn’t Look Like Humans | Supercut
When people think about robots, they usually imagine something like a Boston Dynamics robot, metallic and humanoid. But the robots we'll see in the future might not look like that at all. I mean, if humans are interacting with something on a daily basis, it's probably best not to make it sharp, delicate, and heavy.
How rude.
[Derek] Instead, advanced robots might be made safer if they're soft, flexible, and all kinds of shapes and sizes. So instead of Sonny from I, Robot.
No.
[Derek] Something like Baymax from "Big Hero 6" might be closer to what's in our future.
Balalala.
This is a compilation video of five of my videos on the surprisingly different ways robots can look, and why we build them that way. This is my first time trying out a series, and honestly, it's been a busy time for the Veritasium team. We're working on some exciting things in the background. But in the meantime, we wanted to put this together for all of you. And I also caught up with Dr. Elliot Hawkes, the scientist behind two of these robots, to get an update on how they're progressing and when we can expect to see them in our lives.
Are there more developments happening with the jumping robots?
We have a whole nother project on jumping, and it doesn't even use springs. So I won't give away our our secrets yet. But keep an eye out for that one too, 'cause that's gonna be fun.
So you have another jumping robot which has a totally novel design?
Yep.
And you think it's gonna be better than the one that you had before?
Correct.
And I just wanna shout out our longtime sponsor, Brilliant. They've supported the channel since 2021, and it's great to get to talk about something that I actually use myself. I've had Brilliant on my phone for years, and it just allows me to learn something new every day, instead of, say, doom scrolling. You should make valuable use of your time. So I will tell you more about Brilliant later in the video.
Non-humanoid robots aren't just safer for us to interact with. One of their biggest advantages over traditional robots is that they don't just do things humans already do, but better. Instead, they're specialized to master entirely new abilities, often ones that no human can tackle. This is a robot that can grow to hundreds of times its size, and it can't be stopped by adhesives or spikes. Although it looks kind of simple and cheap, it has dozens of potential applications, including one day maybe saving your life. These robots can be made out of almost any material, but they all follow the same basic principle. Powered by compressed air, they grow from the tip.
That's good.
[Derek] And this allows the robot to pass through tight spaces, and also over sticky surfaces.
Something like a car will get stuck to it. It gets stuck in the wheels. Now if I do the same thing with the vine robot, see the robot is able to extend.
[Derek] It can navigate this curvy and twisted passageway effortlessly, which suggests some of the applications it's well suited for. Now, you might think spikes would be the downfall of an inflatable robot. But even if it's punctured, as long as you have sufficient air pressure, the robot keeps going.
And you might be able to hear it. It's actually leaking now. So I'll have to turn up the pressure. This by itself is not yet a robot. But once we add steering, a camera, some sensors, and maybe some intelligence as to where we're directing it, then we could say it's a robot. So this is sort of the backbone of a robot. This is what allows us to build our type of robots.
[Derek] So where did the idea for this device come from?
I had a vine in my office that was on a shelf, and it was kinda out of the sunlight. And over the course of like a year or so, it slowly grew out this tendril, out and around the edge of the shelf and towards the sunlight. I said, that's a pretty cool thing it just did, right? So you start thinking, well, is there a way you could do that robotically?
[Derek] The solution is really elegant in its simplicity. Just take some airtight tubing and fold it in on itself. It's kinda like a water wiggly, those toys that are really hard to hold. When you inflate it with compressed air, it starts growing out from the tip. And if you want the tube to always bend at a certain spot, you could just tape the tubing on the outside to shorten one of the sides. For example, you could tape it into a helical shape to create a deployable antenna. What about getting them to retract?
Yeah, that's a challenging problem. When you're in a constrained environment, all you really have to do is pull on, we call it the tail, so the material that is passing through the core of the body. You pull on it and it basically outgrows, it just goes back inside itself. Now if you're in a big open area like this and you try pulling on that, instead of inverting, so retracting, it tends to kinda coil up and make a ugly shape.
[Derek] And the engineers have come up with ways to retract the tube to prevent it from buckling using internal rollers. But the tube doesn't have to be the same diameter the whole way along. Here there's actually a much wider section. Think of it like a pillow that's packed into the end of the robot.
Yeah, if you could sit cross-legged on it.
Cross-legged on the table? This sounds super sketchy. So it grows underneath the table just as usual, and then, as the pillow part starts inflating. Is this not good, or is this okay? It can actually lift me up. So my balance is not great, as we can see.
[Elliot] Try standing on it.
Stand on it?
Yeah.
What's amazing is that this doesn't require much pressure above atmospheric. Just a tenth of an atmosphere applied over a large area, like a square meter, can lift something as heavy as a thousand kilograms, all the while remaining soft. Whoa. Woo. That was great. That's the paradoxical thing about pressure. You can get a large overall force with low pressure as long as the area is large enough. What sort of area is that, that pillow there?
It's 600 square inches. So with one PSI, 600 pounds.
Yeah, that's just crazy.
Two PSI, 1,200 pounds.
[Derek] And the whole time, it feels really soft.
Yeah, 'cause there's a couple PSI, right?
[Derek] It's important that the device is still soft so it doesn't hurt anyone.
So you can design these things to have cross-section that changes along its length. So it could be a very small body that could grow into, for example, a collapsed building and potentially lift a large object off someone who's trapped, or maybe in a car crash or something like that. It can apply huge forces with very soft and lightweight, cheap materials.
[Derek] These robots can also be deployed in search and rescue operations by attaching sensors like a camera onto the front.
These robots are actually really hard to stop. So you can take them, grow them into a clutter, potentially a collapsed building or something like that, and they will continue to go somewhere. An alternative is, they're so cheap. I mean, they're basically free. You could grow a hundred of 'em, let's say, into a collapsed building with some sensing on them, and maybe only one of them finds somebody. But I mean, that's a huge success if it does.
[Derek] But how do you keep a camera connected to the front of the robot when it grows out from the tip? Well, one way is to use an end cap, which allows that camera just to stay on the front, pushed from behind by the robot. But there are other mechanisms of attachment. The tiny wireless camera is mounted on an external frame, but this frame interlocks with an internal frame, which is actually inside the pressurized part of the robot body. It's similar to how a roller coaster's wheels go around the track. So this prevents the camera from falling off as the robot grows.
What's really interesting is how the vine robot can be actively steered. They attach artificial muscles to the robot. So the way this muscle works is that if you inflate it, it expands sideways, which leads to it contracting in length.
We don't actually use these much anymore because although it's soft, it's still somewhat stiff. So what we use instead are simply tubes of this ripstop nylon fabric, with the braid oriented at 45 degrees. So in this sense, we just have one single layer of airtight fabric. This is the main robot body here. Then we have three pneumatic muscles connected to it. Now, these three muscles are each connected to their own air supply, connected to regulators over here. As the robot extends from the tip, we can steer it by shortening and lengthening the sides. So just the way your hand works is if I shorten this tendon in my arm, my hand will move this way. Or if I shorten the one on this side, it'll move the other way. So our vine robot, we have these muscles along its side, so as they inflate, they'll turn it one way. Then if I inflate the one on the other side, it'll turn the other way.
[Derek] So the vine robot can fit through tight spaces, it doesn't typically get stuck on anything, and isn't bothered by sharp objects. And once you attach that camera on the front, it's ideal for things like archeology. The robot was actually taken to Peru to investigate some very narrow shafts.
So we were looking at this archeological site that was built somewhere between 1,500 and 500 BC in the Andes Mountains of Peru, and it was an ancient temple that had all these underground spaces. And part of what the archeologists were doing was trying to understand what the spaces were for, and what the people who built them were trying to do with them. So part of that was unknown. But there were these giant rooms that they called galleries, and then there were these small ducts, or tunnels that were offshoots of these rooms, and they wanted to know where these ducts led, but they were too small for a person to go in. So we were able to successfully use the vine robot to explore three of the tunnels that couldn't have been seen through other means, which was super exciting. And we got video inside the entire tunnels and gave it to the archeology team.
[Derek] There's an application where I feel like this solution is just so obvious I wonder why it didn't exist before.
Intubation is literally the process of putting a tube into a patient. The purpose is to breathe for the patient when the patient isn't breathing. And so traditionally, a highly-trained medical professional would take their laryngoscope, come above the patient. And once they see the trachea, you start to pass your tube down inside. I'm almost there. I can see the light. So if you can see right now, I just got it in to the trachea.
Oh, yeah.
Right there. And it took me a couple minutes, and I was really kinda wrenching on this patient here. So if there's somebody who's not breathing, every second counts.
[Derek] But by using a miniature version of this vine robot, researchers are hoping to make intubation faster and safer.
Somebody like me with no training could pretty simply insert this device, aim towards the nose, and just like that, if you can see, we've already intubated, and all it took was a little bit of pressurization. Just like that.
[Derek] It almost looks like a sort of a party favor.
Yeah, right? It's like a, this reminds me a lot of those inflatable kinda like Play-Doh structures that you see at car lots.
[Derek] How does it know to go down the right tube?
Yeah, so that's one of the kind of cool things about soft robotics, is the robot is quite compliant, and we see that in a lot of these demos. They can squish, they can bend. And so, how we've designed it is that the main robot grows down into the esophagus, and then we have this side branch that heads towards the trachea, and it's quite flexible. And so, it basically finds the opening. So it's a really neat example of kind of a passive intelligence, mechanical intelligence, some people call it, where it can find its path even if we don't know exactly the shape beforehand.
[Derek] Have you tried this on a real person yet?
Not on a real person, but we've actually tried this in a cadaver lab, and we've shown that we can move from this nice idealized version to an actual in vivo situation and successfully intubate a patient.
[Derek] There's another application, which is burrowing into sand or soil. When you blow compressed air into something like sand, it fluidizes, it becomes like a liquid, and that can allow the vine robot to grow into granular materials like sand.
If you've ever been to the beach and you try to stick your umbrella pole into the ground, it's fairly difficult.
I'll try to push that probe down into the sand, no fluidization. Yeah, it feels like it sort of gets wedged in there.
So now I'll turn on the air.
Oh yeah, you can feel it immediately. Oh, wow. Yeah, that's a lot.
So what we've done here is essentially, we just blow a jet of air out the front of the robot, and that loosens up the sand enough to reduce the force of the sand so that the robot, just by tip extension, can make its way through. (air whooshing) (rousing music)
This makes vine robots an attractive option for NASA when they look for ways to study the surfaces of other planets. Recently on Mars, they tried to have a burrowing robot, but it got stuck. Could you do it better basically with this?
Yeah, that's a good question. So the Mars InSight mission, they have this heat probe. The idea there was to be able to sort of hammer its way down into the core and then place a sensor that could detect the temperature of Mars. However, the problem they ran into there is that it turned out the material that they put it in was more cohesive than they expected. Inside the robot, something would wind up and pound it down, wind up and pounded down. But it turned out there wasn't enough friction between the probe and the sand. So what was really happening was it would wind up, pound down, wind up, pound down, wind up, pound down. So it'd never actually go anywhere. The advantage of something like this, like tip extension, is you'd have your base, you start at the surface, and you just keep extending your way down. You're not necessarily relying on the interaction with what is surrounding it to make it work.
[Derek] What amazes me about vine robots is how a plant inspired this simple, elegant design. It's so easy in fact that you could build one yourself in as little as a minute. There are instructions online that I'll link to. But from that basic design have come a huge variety of robots, with different applications from archeology to search and rescue, or intubation to space exploration. And what else can you think of to do with it?
I've actually gotten a lot of emails from viewers about crazy ideas that we hadn't thought of. So keep 'em coming. We love to hear your ideas.
Are there any ideas that you can share with us that are like, oh, like that's actually really cool?
One idea was for clearing mines, landmines. So the idea was that you'd actually run a vine robot through the field and then detonate the landmines, and basically make a path that civilians could walk through the field. So I thought that was kind of a cool idea.
Do you think they'd create enough pressure to trigger these things?
I think the idea was they were gonna actually put explosives in the vine to detonate the, yeah, the land mines themselves.
I mean, I'm also thinking, now that you mention that, like I'm thinking you could put metal detector type sensors on the vine robot.
Absolutely.
So spread them across the field and they'll pick up where the mines are.
I'll give you another crazy example. Another one was for space applications of docking, so two spacecraft docking together, you have to make an airtight seal. And so, the idea was, well, maybe you could use a vine robot to do this.
Some sort of like airlock or something going out and then.
So basically, yeah, you can imagine the two tubes coming together, not sealed, and then the vine robot growing through and basically making the seal.
Are there any updates about the vine robot? Like have the medical trials gotten anywhere?
So we just did a trial with emergency medical practitioners using our device. We gave 'em five minutes of training, we gave 'em the device, and they were 90-ish percent successful in intubating, in very rapid, something like 20 seconds.
Oh, wow.
The nice thing about our device is that if it fails, it fails in 20 seconds. And so, it doesn't take three minutes to attempt and then realize you didn't get it. I think something really nice about that is that it is so rapid and easy that even if it does fail, you get another shot.
Yeah, I mean, that one seems like it's so close to actually having a big real world application. I mean, how common are intubations?
So intubations in the OR, so operating room are quite common. I think like 15 million a year. That's probably not our initial target 'cause those are very reliable, like 99% plus. The fundamental problem is that the tools are designed for those doctors in those scenarios. And what happens is those similar tools get put out into the field for an ambulance and a paramedic is trying to do an intubation, but they're doing it maybe in the dark, with someone in a poor body position, where there's blood in the mouth. So our device takes that required skill out and basically lets the vine robot find the way into the trachea. And so, pre-hospital, there's around a million intubations a year. And we think many intubations that aren't even attempted because the tools just aren't there. And then our kind of long shot is eventually, there's AEDs, the defibrillators everywhere. One issue there is there's not a way to help the person breathe. Possibly if we can make this thing so simple, it could basically be packaged with a AED, where you could both get the heart going, and you could intubate and get the oxygen in. But I think we're close. It's pretty easy. You'll have to come back and get another video. We'll let you intubate a cadaver. That'll be fun.
So we've kind of answered this question, but are these robots, vine robots still being worked on?
We have a project right now on anchoring, especially we're interested in. So if you think of a plant root. If you ever try to pull out, I don't know, like a small shrub or something, it's incredibly hard, right? Like you're looking at this, it's got like a half an inch stem and you're pulling on it and pulling on it, it's like hundreds of pounds of force. So you're like how is this possible? And I think one of the coolest things is that a hundred pounds of anchoring force was created with almost no reaction force initially. There was like a seed that slowly grew down into the ground.
It's like all these little tendrils, and I'm imagining the friction sums over all of those.
So basically, when you're trying to go into the soil, the thing resisting you is the surface area of the tip. That's what you're pushing in. And then what's giving you the anchoring force is the surface area on the sides. And so, you can imagine, if you clump 'em all together, the area in the tips doesn't change, but the surface area of the sides went down. So you basically wanna split 'em up into as many, practically. So anyway, we're using that concept now to make these anchors, and we're working with NASA now one as well. And so, it's like this deployable anchor, it's very light. You could throw it somewhere, or just drop it, and then the roots grow down, there's four roots that grow down into the ground, and it was something like a hundred Newtons of force to pull it out.
Yeah, it sounds like a very sci-fi type thing, where you could like throw the root pack down and it just like phmf, and then the roots roots grow out and you're like, oh yeah, the anchor is locked in.
Yep, absolutely, absolutely.
Yeah, that's amazing. After seeing the unstoppable robot, we returned to Elliot's lab a few years later to see a robot that has conquered a totally different specialty, the art of jumping. This tiny robot weighs less than a tennis ball, and can jump higher than anything in the world. In the competitive world of jumping robots, the previous record was 3.7 meters, enough to leap a single story building. This jumper can reach 31 meters, higher than a 10-story building. It could jump all the way from the Statue of Liberty's feet up to eye level.
For something to count as a jump, it must satisfy two criteria. First, motion must be created by pushing off the ground. So a quadcopter doesn't count, because it pushes off the air. And second, no mass can be lost. So rockets constantly ejecting burnt fuel are not jumping, and neither is an arrow launched from a bow. The bow would have to come with the arrow for it to count as a jump. Many animals jump, from sand fleas, to grasshoppers, to kangaroos, and they launch their bodies into the air with a single stroke of their muscles. The amount of energy delivered in that single stroke determines the jump height. So if you wanna jump higher, you have to maximize the strength of the muscle. The best jumper in the animal kingdom is the galago, or bush baby, and that's because 30% of their entire muscle mass is dedicated to jumping. This allows the squirrel-sized primate to jump over two meters from a standstill.
It has like very small arms and upper body, and has just like huge jumping legs. It doesn't have better muscles or anything, it just has more of them.
[Derek] There are some clever jumping toys.
I feel like, there we go. Oh!
I used to play with these poppers as a kid. And when you deform a popper, you store energy in its deformed shape. Effectively it becomes a spring. And then, just like an animal, in one stroke, it applies a large force to the ground, launching itself into the air. All elastic jumpers follow the same principle of storing energy in a spring and releasing that energy in a single stroke to jump. But none of the jumping toys we had could compare to this tiny robot. Of all the things that I have ever tried to film, this is the most challenging. Because it's so small, it accelerates rapidly and travels a huge distance on each jump. Each takeoff happened faster than we could even register.
Now, jumping might sound like a niche skill, but engineered jumpers would be perfect for exploring other worlds, particularly where the atmosphere is thin or non-existent. On the moon, with one sixth the gravity of earth, this robot would be able to leap 125 meters high and half a kilometer forward. Rovers may struggle with steep cliffs and deep craters, but jumpers could hop in and out, fetching samples to bring back to the rover. And you don't lose much energy when jumping. So if you could store the kinetic energy back in the spring on landing, the efficiency could be near perfect. The team has already started to build an entire fleet of jumping robots. Some of them can right themselves after landing, so they can take off again right away. Others are steerable. They have three adjustable legs that allow the jumper to launch in any direction.
Essentially what we've done is we've added three additional legs that don't store energy but rather, allow it to form a tripod sort of that allows it to point a direction and launch in that direction.
[Derek] But how does this jumping mechanism work? Well, the main structure consists of four pieces of carbon fiber, bound together by elastic bands. Together, they create a spring that stores all the energy needed for the jump. At the top of the robot is a small motor. A string wrapped around the axle is connected to the bottom of the robot. So when the motor is turned on, it winds up the string, compressing the robot, and this stores energy in the carbon fiber and rubber bands. After about a minute and a half, the structure reaches maximum compression. How do you know when to put it down?
Basically once the bottom there sticks inward and it can stand up, right now it would roll over, then you can put it down.
Got it.
So as soon as you can.
[Derek] And at this point, a trigger releases the latch that's holding the string on the axle. So all the string unspools all at once, and the energy stored in the spring is released.
Yeah.
[Derek] The jumper goes from a standstill to over a hundred kilometers an hour in only nine milliseconds. That gives an acceleration of over 300 Gs. That would be enough to kill basically any living creature.
Watch out, watch out, watch out.
But how does it jump so much higher than everything else, nearly 10 times higher than the previous record holder? Well, this jumper has three special design features. First, the jumper is incredibly light, at just 30 grams. It achieves this weight by employing a tiny motor and battery. Plus, its entire structure, made of lightweight carbon fiber and rubber, doubles as the spring. Per unit mass, natural latex rubber can store more energy than nearly any other elastic material, 7,000 joules per kilogram. And the design of the spring makes it ideal for its purpose. Initially they tried using only rubber bands connected to hinged aluminum rods. But with this design, when compressing it, the force rises to a peak and then decreases. Just feels like it all of a sudden got a lot easier to pull.
Another design with only carbon fiber slats requires a lot of force to get started, and then it increases linearly after that. There is more and more force required to do this. The ultimate design is a hybrid of these two approaches. The benefit being, its force profile is almost flat over the entire range of compression. It feels like that needs a lot of force. And now it feels pretty steady with the amount of force that I need to apply. Therefore, it provides double the energy storage of a typical spring, where force is proportional to displacement. The researchers argue this is the most efficient spring ever made.
Sometimes a string'll snap. It's not always consistent that it releases when it's supposed to.
Oof!
The string cut it. Lemme go restring it. I'll be right back.
[Derek] All right. You'd probably expect that lighter would always be better with a jumper, especially if the added weight is simply dead weight rather than anything useful, like a spring or a motor.
So we're adding basically a chunk of steel to our jumper and it's gonna jump higher, and the key is that we're adding it to the top. You want your body, the part that's moving, to weigh at least as much as the foot. When your body's lighter, it's basically just collision, this energy transfer is very inefficient and you don't jump very high.
But the real secret to how this jumper can achieve such heights is through something the researchers call work multiplication. Unlike an animal, which can only jump using a single stroke of its muscle, an engineered jumper can store up the energy from many strokes, or in this case, many revolutions of its motor, and that's how the motor can be so small. It doesn't have to deliver the energy all at once. It builds it up gradually over a few minutes. So the trade-off is kind of like time for energy.
[Elliot] Exactly.
[Derek] And this is possible because there is a latch under tension preventing the spring from unspooling until the robot is fully compressed. Interestingly, biological organisms do use latches. For example, the sand flea, which can jump incredibly high for its body size.
It has a muscle that is attached, let's say, right here, is right inside of the pivot point. So as it contracts that muscle, the leg doesn't extend, right? It's actually closing it more. But then it has a second muscle that pulls it out. It's going to shift this muscle ever so slightly outside the pivot point.
That's wild. So there's these two muscles that are working.
So here's your big power muscle. Here's your trigger muscle. It's a torque reversal mechanism. And then all of a sudden, it shoots.
[Derek] But even though the biological world has latches, no organism has developed work multiplication for a jump from standstill, at least not internally. Spider monkeys have been observed pulling back a branch hand over hand using multiple muscle strokes stored in the bend of the branch to catapult themselves forward. There's a spider that shoots out a silky string, which they pull back multiple times in order to slingshot themselves to another location. So it's like slingshotting itself?
Yes, they're called the slingshot spider.
Now, I tried jumping in moon boots to see if they would help me go higher. (remarks drowned out by background noise)
Okay. Oof. And it certainly felt like they did, but Elliot pointed out that from a standing start, they don't actually help much.
Kind of build it, build it, build it, and then go.
Okay. Only if you jump a few times before can you store up some of the previous jumps energy in the elastic bands, and then that energy helps launch you higher on the following jump. For years, engineered jumping was developed to mimic biological jumping. But with work multiplication, it gained an advantage. If you can generate a large burst of energy simply by running a motor for a long time, the power of the motor is no longer the limiting factor, the spring is. So you can focus on making the most powerful spring possible. This jumper has nearly maximized the achievable height with this spring. Assuming an infinitely light motor with infinite time to wind up, the highest possible jump with this compression spring is only around 19% higher than what they've achieved.
If you want to incorporate air resistance and play with aerodynamics, another way to send the jumper higher is to make it 10 times isometrically larger, leading to a 15 to 20% higher jump.
So we're in kind of an intermediate scale where we still are getting hit by air drag, but it's not as bad as the flea. If we went 10 times bigger, we could actually avoid air drag completely.
This works since if the jumper is scaled up 10 times on all sides, the cross-sectional area increases by a hundred, which increases the drag force, but the jumper's mass increases by a thousand. So it has way more inertia, meaning the drag force affects it less.
The entire concept of work multiplication could bring robots to the next level. Currently, motors and robots have to be relatively small so they remain portable. But the simple principle of building up the energy from multiple turns of a motor over time would allow robots to store and then release huge amounts of energy, and set some world records in the process.
When we visited you, we were looking at 110 feet and it was the record holder. My question, is that still the record as far as you know?
It is. I also challenge all the viewers to beat it, because it is beatable. So I hope someone in the next few years will beat it, and if not, we'll beat our own record, because it's beatable. I will say that much.
Are there more happening with the jumping robots?
We have a whole nother project on jumping, which we also think will beat it, and it doesn't even use spring. So I won't give away our secrets yet. But keep an eye out for that one too, 'cause that's gonna be fun.
So you have another jumping robot, which has a totally novel design?
Yep.
And you think it's gonna be better than the one that you had before?
Correct.
Wow, that's extreme. And in terms of making these things applicable?
Yeah, so I mean, we do have a project with NASA. I will say, our stuff with NASA moves pretty slow, just in terms of what they really care about is getting something really, really reliable. If they're gonna send it to the moon, it can't mess up. And so, that's a slow going process. But I just think that's still a goal, is to get it to the moon.
It would be like the next Mars helicopter or something.
I think that's a really good analogy. So the Mars helicopter has just been doing awesome, basically being this little scout that can just go out and get some nice views of where the rover might need to go, or maybe getting some samples. You can't have a helicopter on the moon, but you can jump. And we think we can get pretty similar performance in terms of height and stuff like that with jumping on the moon compared to the helicopter. So yeah, no, that's a great analogy.
Are there any interesting emails that came about from the jumping robot video?
So a lot of people want to make one, and I will say, it's really hard, and I keep writing that it's really hard. Every piece in that robot was... They talk about safety factors in engineering where you're supposed to have a good safety factor, and we had no safety factors in anything. And so, pretty much everything was near its failure limit. And so, that just made it incredibly hard. But what we're trying to do actually right now is put together a tutorial to make like a, you know, it's not gonna jump a hundred feet, but it'll jump maybe 20 or 30 feet. Stay tuned for that, 'cause I think that would be a fun way to get people to build one themselves and try it out. I will say though that if you really push the limits, what we're trying to do, wear your safety glasses and your gloves, and all of that. Because the number of shards of carbon fiber I've gotten into my fingers.
Oh, no. Brutal. What is the biggest thing you learned building the jumping robot?
How many things can go wrong when you're trying to build something really cool. 'Cause this was years and years of failing, over and over, and over and over again. So I think something that sticks out to me is that it takes a lot of failure to get a success like that.
Would anyone ever say to you, well, why did it take that many years? Could you not have modeled out the springs, like done some simulations of it before you actually build the thing?
Well, no, and I should mention too that it's not like we didn't do any modeling or simulation. That's all part of our cycle too. But the problem is, I mean, we didn't even know what shape. We went through so many different configurations of robots. Like some were kind of this ball shape, but we had other robots that were more stick like, rubber band based. It's like this crazy search over a huge space, and comparing all kinds of different trade offs. I think that's a reasonable amount of time to make a world record, but that's just me.
The jumper robot can jump so high precisely because it's built of rubber and carbon fiber, with a tiny body and massive legs. Its whole body is designed for one physical purpose. The specialization of robots is also natural in academic research, since it's easier for scientists to isolate and understand one ability. These may later be included into a single more complex robot. People expect sort of the Boston Dynamics, humanoid type robots. Why have you investigated these sort of other very strange looking type robots?
The thing that maybe unifies all of them is, they're mostly about mechanical design, my robots, and less so about controls, and vision, and AI, and that side. And so, often we think of what we do are kind of adjacent to core robots. But our lab isn't as focused on traditional robotics just because I find more joy myself in mechanical design. So that's what I like to do.
Do you have a personal favorite robot?
I can't play favorites with my robots.
I know, it's like picking amongst your children. Like it's just impossible.
I love all my robots. I love all my robots. I will say though that the jumper, there was a certain, it took a lot of learning from our failures and revising it. And so, when we finally got that one working, I think that was really satisfying.
From your perspective, how did our collaboration come to be in the first place?
Oh, wow. Okay. That's a fun story. So you sent me an email, I don't know how many years ago, maybe five years ago, something like this, and I ignored it. And I went into lab one day, I just mentioned to my student, I was like, oh, some guy emailed about making a video. And he is like, "Yeah, who was it? Maybe I could help out or something." So I forwarded it to him, he's like, "Do you know who this is?" So then we responded. And we appreciated the effort you put in to the details in getting the story right. I forget how long we had this call. I had this call with Emily, it was maybe four hours or something. We went through the details of jumping theory. 'Cause she wanna get everything just right for that jumping video. As an academic, we care about that stuff and we want it to be right.
But the most specialized perfected match between a robot's build in its abilities comes from one competition that's been refining this for nearly 50 years. Micromouse is the oldest robotics competition in the world. It's like the Formula 1 of robotics, but you have to see their speed to believe it. This tiny robot mouse can finish this maze in just six seconds. (rousing music)
Every year around the world, people compete in the oldest robotics race. The goal is simple: get to the end of the maze as fast as possible.
Person who came second (spectators cheering) lost by 20 milliseconds.
[Derek] But competition has grown fierce.
When somebody saw my design, they said, you're crazy.
[Derek] Why is there so much tension? What's riding on it? Honor?
Honor. (spectators applauding)
[Derek] In 1952, mathematician Claude Shannon constructed an electronic mouse named Theseus that could solve a maze. The trick to making the mouse intelligent was hidden in a computer built into the maze itself, made of telephone relay switches. The mouse was just a magnet on wheels essentially, following an electromagnet controlled by the position of the relay switches.
[Claude] He is now exploring the maze using a rather involved strategy of trial and error. As he finds the correct path, he registers the information in his memory. Later, I can put him down in any part of the maze that he's already explored and he'll be able to go directly to the goal without making a single false turn.
Theseus is often referred to as one of the first examples of machine learning. A director at Google recently said that it inspired the whole field of AI. 25 years later, editors at the Institute of Electrical and Electronics Engineers, or IEEE, caught wind of a contest for electronic mice, or le mouse electronique, as they had heard. They were ecstatic. Were these the successors Theseus? But something had been lost in translation. These mice were just batteries in cases, not robots capable of intelligent behavior. But the misunderstanding stuck with them, and they wondered, why couldn't we hold that competition ourselves? In 1977, the announcement for IEEE's amazing Micromouse Maze Contest attracted over 6,000 entrants. But the number of successful competitors dwindled rapidly. Eventually, just 15 entrants reached the finals in 1979. But by this point, the contest had garnered enough public interest to be broadcast nationwide on the evening news. And just like the rumor that inspired the competition, Micromouse began to spread across the world.
♪ Micromouse for the taking ♪
♪ Micromouse, it's here and now ♪
♪ Take a chance and start creating ♪
♪ Micromouse will show you how ♪
The Micromouse.
[All] Micromouse! (upbeat music)
Even people in the top two or three, you can see them trying to set their mice up, and they can barely find the buttons to press because it's absolutely nerve wracking. (suspenseful music) It doesn't matter what it was. It could be horse racing, it could be motor racing, it could be mouse racing. If you have a shred of competitiveness in you, you wanna win, right?
[Derek] Just like a real mouse, a micromouse has to be fully autonomous, no internet connection, no GPS or remote control, and no nudging it to help it get unstuck. It has to fit all its computing, motors, sensors, and power supply in a frame no longer or wider than 25 centimeters. There isn't a limit on the height of the mouse, but the rules don't allow climbing, flight, or any forms of combustion. So rocket propulsion for example, is out of the equation. (whistle being blown) The maze itself is a square about three meters on each side, subdivided by walls into corridors only 18 centimeters across. And in 2009, the half-size Micromouse category was introduced, with mice smaller than 12 and a half centimeters per side, and paths just nine centimeters across. The final layout of the maze is only revealed at the start of each competition, after which competitors are not allowed to change the code in their mice.
(spectators applauding) The big three competitions, all Japan, Taiwan, and USA's APEC, usually limit the time mice get in the maze to seven or 10 minutes, and mice are only allowed five runs from the start to the goal.
So if you spend a lot of time searching, that's a penalty.
Makes sense. So the strategy for most micromice is to spend their first run carefully learning the maze and looking for the best path to the goal, while not wasting too much time. Then they use their remaining tries to sprint down that path for the fastest runtime possible.
(spectators applauding) Solving a maze may sound simple enough, though it's important to remember that with only a few infrared sensors for eyes, the view from inside the maze is a lot less clear than what we see from above. Still, you can solve a maze with your eyes closed. If you just put one hand along one wall, you will eventually reach the end of most common mazes, and that's exactly what some initial Micromouse competitors realized too. And after a simple wall following mouse took home gold in the first finals, the goal of the maze was moved away from the edges and freestanding walls were added, which would leave a simple wall following mouse searching forever.
Your next instinct might be to run through the maze taking note of every fork in the road. Whenever you reach a dead end or a loop, you can go back to the last intersection and try a different path. If your last left turn got you nowhere, you'd come back to that intersection and go right instead. You can think of this strategy as the one a headstrong mouse might use, running as deep into the maze as it can and turning back only when it can't go any further. This search strategy, known as depth-first search, will eventually get the mouse to the goal. The problem is, it might not be the shortest route, because the mouse only turns back when it needs to, so it may have missed a shortcut that it never tried.
The sibling to this search algorithm, breath-first search, would find the shortest path. With this strategy, the mouse runs down one branch of an intersection until it reaches the next one, and then it goes back to check the path it skipped before moving on to the next layer of intersections. So the mouse checks every option it reaches, but all that backtracking means that it's rerunning paths dozens of times. At this point, even searching the whole maze often takes less time. So why not just do that?
A meticulous mouse could search all 256 cells of the maze, testing every turn and corner to ensure it has definitely found the shortest path. But searching so thoroughly isn't necessary either. Instead, the most popular Micromouse strategy is different from all of these techniques. It's a search algorithm known as flood-fill. This mouse's plan is to make optimistic journeys through the maze, so optimistic in fact, that on their first journey their map of the maze doesn't have any walls at all. They simply draw the shortest path to the goal and go. When their optimistic plan inevitably hits a wall that wasn't on their map, they simply mark it down and update their new shortest path to the goal. Running, updating, running, updating, always bee lining for the goal.
Under the hood of the algorithm, what the micromouse is marking on their map is the distance from every square in the maze to the goal. To travel optimistically, the mouse follows the trail of decreasing numbers down to zero. Whenever they hit a wall, they update the numbers on their map to reflect the new shortest distance to the goal. This strategy of following the numerical path of least resistance gives the flood-fill algorithm its name. The process resembles flooding the maze with water and updating values based on the flow.
Once the mouse reaches the goal, it can smooth out the path it took and get a solution to the maze. However, it may look back and imagine an even shorter uncharted path it could've taken. The mouse might not be satisfied that it's found the shortest path just yet. While this algorithm isn't guaranteed to find the best path on the first pass, it takes advantage of the fact that micromice need to return to the start to begin their next run. So if the mouse treats its return as a new journey, it can use the return trip to search the maze as well. Between these two attempts, both optimized to find the shortest path from start to finish, it's extremely likely that the mouse will discover it, and the mouse will have done it efficiently, often leaving irrelevant areas of the maze entirely untouched. Flood-fill offers both an intelligent and practical way for micromice to find the shortest path through the maze.
Once there was a clear strategy to find the shortest path, and once the microcontrollers and sensors required to implement it became common, some people believed Micromouse had run its course. As a paper published in IEEE put it, "At the end of the 1980s, the Micromouse Contest had outlived itself. The problem was solved and did not provide any new challenges."
In the 2017 all-Japan Micromouse competition, both the bronze and silver placing mice found the shortest path to the goal. And once they did, they were able to zip along it as quick as 7.4 seconds. (spectators applauding) But Masakazu Utsunomiya's winning mouse, Red Comet, did something entirely different. This is the shortest path to the goal, the one that everyone took. This is the path that Red Comet took. It's a full five and a half meters longer. That's because micromice aren't actually searching for the shortest path, they're searching for the fastest path, and Red Comet's search algorithm figured out that this path had fewer turns to slow it down. So even though the path was longer, it could end up being faster. So it took that risk. (spectators applauding)
It won by 131 milliseconds. Differing routes at competition are now more common than not, and even just getting to the goal remains difficult, whether due to a mysterious algorithm or a quirk of the physical maze.
[Announcer] On the corner, it's a little bit like a, whoa.
[Derek] Micromice don't always behave as you'd expect. Micromouse is far from solved, because it's not just a software problem or a hardware problem, it's both, it's a robotics problem. Red Comet didn't win because it had a better search algorithm, or because it had faster motors. Its cleverness came from how the brains and body of the mouse interacted together.
So it turns out solving the maze is not the problem, it never was the problem, right? But it's actually about navigation, and it's about going fast.
Every year, the robots get smaller, faster, lighter. There is still plenty of innovation left. And there's a small group of devotees in Japan busy building quarter-sized micromouse, which would sit on a quarter.
- [Derek] Nearly 50 years on, Micromouse is bigger than ever. (spectators applauding)
Competitions have appeared solved at first glance before. The high jump was an Olympic sport since 1896, with competitors refining their jumps using variations like the scissor, the western roll, and the straddle over the decades with diminishing returns. But once foam padding became standard in competition, Dick Fosbury rewrote the sport in 1968 by becoming the first Olympian to jump over the pole backwards.
Now almost every high jumper does what's known as the Fosbury Flop. If Micromouse had indeed stopped in the 1980s, the competition would've missed its own Fosbury flops, two innovations that completely changed how micromice ran. After all, a lot can change in a sport where competitors can solder on any upgrade they can imagine.
The first Fosbury flop was one of the earliest innovations in Micromouse, and had nothing to do with technology. It was simply a way of thinking outside the box, or rather, cutting through the box. Every mouse used to turn corners like this. But everything changed with the mouse, Mitee 3.
The Mitee mouse three implemented diagonals for the first time. (spectators cheering) And that turned out to be a much better idea than we really thought. And because it's cool, maze designers often put diagonals into the maze now. So, you know, you could end up with a maze where it never comes up, but most of the time, it's actually a benefit.
[Derek] In order to pull off diagonals, the chassis of the mouse had to be reduced to less than 11 centimeters wide, or just five centimeters for half-size Micromouse. The sensors and software of the mouse had to change too. When you're running between parallel walls, all you have to do is maintain an equal distance between your left and right infrared readings. But a diagonal requires an entirely new algorithm, one that essentially guides the mouse as if it had blinders on.
Normally if you're going along the side of a wall or something like that, most of the time, you can see the wall all the time. And so, that helps you to guide yourself, and you know when you're getting off. But in the diagonal situation, you just see these walls coming at you.
[Derek] And if you veer even a tiny bit off course, snagging a corner is a lot less forgiving than sliding against a wall. Diagonals are still one of the biggest sources of crashes in competition today. But in exchange, a jagged path of turns transforms into one narrow straightaway.
Whoa, whoa! (spectators applauding) (spectators applauding) (spectators cheering)
[Derek] These days, nearly every competitive micromouse is designed to take this risk. Cutting diagonals opened up room for even more ideas.
Around the same time, mice were applying similar strategies to turning. Instead of stopping and pivoting through two right turns, a mouse could sweep around in a single U-turn motion. And once the possibility of diagonals were added, the total number of possible turns opened up exponentially. The maze was no longer just a grid of square hallways. With so many more options to weigh, figuring out the best path became more complex than ever. But the payoff was dramatic. What was once a series of stops and starts could now be a single fluid snaking motion.
(spectators applauding) Available technology was getting upgrades over the years as well. Tall and unwieldy arms that were used to find walls were replaced by a smaller array of infrared sensors onboard the mouse. Precise stepper motors were traded in for continuous DC motors and encoders.
The DC motors give you more power for less size and weight. And so, we were interested in doing that. So then you have to have a servo, you have to actually have feedback on the motor to make it do the right thing.
[Derek] Gyroscopes added an extra sense of orientation. It's like a compass.
Absolutely.
You had this thing with you.
They came about 'cause of mobile phones, really. So the technology provides people with things which weren't there before.
All of the turning is done based off the gyro, rather than counting pulses off the wheels, 'cause it's much more reliable.
- [Derek] But even with all the mechanical upgrades, the biggest physical issue for micromice went unaddressed for decades. One thing you'll see almost every competitor holding is a roll of tape. Once you know to look for it, you'll see it everywhere. This tape isn't for repairs or reattaching fallen parts. It's to gather specks of dust off the wheels in between rounds.
If you wanna turn while driving fast, you need centripetal force to accelerate you into the turn. And the faster you're moving, the more force you need to keep you on the track. The only centripetal force for a car turning on flat ground is friction, which is determined by two things, the road pushing up the weight of the car, or the normal force, multiplied by the static coefficient of friction, which is the friction of the interface between the tire and road surface. This is why racetracks have banked turns. The steep angles help cars turn with less friction, because part of the normal force itself now points in to contribute to the centripetal force required. If the bank turn is steep enough, cars can actually make the turn without any friction at all. The inward component of the normal force alone is enough to provide the centripetal force required to stay on track.
Micromice are no different, and they don't have banked turns to help. As they got faster and faster, by the early 2000s, their limiting factor was no longer speed, but control of that speed. They had to set their center of gravity low and slow down during turns to avoid slipping into a wall or flipping over. But unlike race cars, there wasn't anything in the rules to stop Micromouse competitors from solving this problem by engineering an entirely new mechanism.
Micromouse's second Fosbury flop was almost considered a gimmick when the mouse Mokomo08 first used it in competition. You might be staring at the video to try to see it, but you won't. Instead, it's something you'll hear. That isn't the mouse revving its engines. It's spinning up a propeller. And while flying over the walls is against the rules, there's nothing in the rules against a mouse vacuuming itself to the ground to prevent slipping.
Dave Otten was the first person I saw put a fan on a mouse, but he used a ducted fan, and I think he was really looking at kind of reaction force, blowing the thing down. He had a skirt around, but it was not terribly effective. He'll forgive for saying so. The idea is to let as little air in as possible. And like your vacuum cleaner, when you block your vacuum cleaner, the motor unloads and speeds up, and so the current drops. But if you let too much air in, the current's very high. And these are just quadcopter motors, and they draw a lot of current.
[Derek] At the scale of Micromouse, a vacuum fan, often just built from handheld drone parts, is enough to generate a downward force five times the mouse's weight.
Wow, okay. That's impressive. So how much does the car actually weigh?
About 130 grams. And if you listen, I dunno if you'll get it on your microphone, but.
Oh, yeah.
You hear the motors slow down, loads up.
[Derek] With that much friction, micromice today can turn corners with a centripetal acceleration approaching 6 Gs. That's the same as F1 cars. Once nearly everyone equipped fans, the added control allowed builders to push the speed limit on micromice.
When it's allowed to, it will out accelerate a Tesla Roadster, but not for very far.
[Derek] And they can zip along at up to seven meters per second, faster than most people can run. (spectators applauding) (spectators cheering) (spectators applauding)
Every one of the features now standard on the modern micromouse was once an experiment, and the next Fosbury flop might not be far off. The first four-wheeled micromouse to win the all-Japan competition did so in 1988, but it would take another 22 years of the winning mouse growing and losing appendages before four wheeled mice became the norm. With micromice still experimenting in six and eight wheeled designs, omnidirectional movement, and even computer vision, who knows what the next paradigm shift will be.
[Announcer] Your time on the maze actually begins only when you leave the start square. So he's not penalized for any of this time.
But if you wanna get started with Micromouse, you don't need to worry about wheel count, or vacuum fans, or even diagonals.
It is to my mind, the perfect combination of all the major disciplines that you need for robotics, and engineering, and programming, embedded systems, all wrapped up in one accessible bundle that you can do in your living room, and you don't need a laboratory to run it. You come along because you're curious. And then you think, I could do that, that doesn't look so hard. And then you're doomed, really. If it sucks you in, it turns into quite the journey. (spectators cheering) (spectators applauding)
At its core, Micromouse is just about a mouse trying to solve a maze. Though nearly 50 years later, it's a simple problem that's a good reminder, there is no such thing as a simple problem.
♪ Micromouse for the taking ♪
♪ Micromouse, it's here and now ♪
♪ Take a chance, start creating ♪
♪ Micromouse will show you how ♪
A humanoid robot built for all the same tasks a human does sacrifices specialization in any one skill in order to be a generalist. But if it does all tasks semi well and these tasks are what humans are already doing, well, then those robots are just overlapping with us, copying our capacities rather than expanding them.
So robots are perhaps likely to enter our lives not as multipurpose humanoids, but rather as precise tools we can pick and choose. Instead of one Swiss army knife robot, you'd end up with something like a personalized toolbox of specialized robots. Big futuristic questions like how to bring robots into our daily lives requires all sorts of practical and creative skills. But perhaps the most important one is actually something that anyone can build, problem solving.
If you wanna hone your own ability to problem solve, you can get started on that right now, for free, with today's sponsor, Brilliant. Brilliant will make you a better critical thinker, while helping you build real skills in everything from technology and programming to math, data science, and whatever you're curious about.
On Brilliant, you'll learn through discovery by trying things yourself. And you'll not only gain knowledge of key concepts, but learn to apply them to real world situations, all while building your intuition. This gives you the tools to solve whatever problems come your way. You know, we've been partnering with Brilliant for years, and I still find new things to learn all the time, like this course on how to maximize the value of an electric car. Brilliant has thousands of interactive lessons, and because each one is bite-sized, you can do it in just minutes. So no matter where you are, you can always be building a quicker, sharper mind right on your phone.
To try everything Brilliant has to offer for free for 30 days, visit brilliant.org/veritasium, or you can scan this QR code, or click that link down in the description. You'll also get 20% off an annual premium subscription. So I wanna thank Brilliant for sponsoring this part of the video, and now, dive into a surprising trait we're starting to build into robots. Elliot's vine robot and jumping robot are just two cases where the robots that might save our lives or explore new planets don't look much like traditional robots at all. But soft robots in particular are an entire field of study.
So why are so many researchers trying to build the robots of the future with soft materials? For one, trading out fragmented metal frames for single flexible bodies might be how we make robots more reliable and precise. These bendy gear boards are so predictable that they were commissioned by the US government to secure nuclear weapons, ensuring that no random motions could accidentally set them off. But predictability is just one of eight reasons that machines that bend are better.
What do this satellite thruster, plastic tool, and micro mechanical switch have in common? Well, they all contain components that bend, so-called compliant mechanisms.
So it's always been considered to be bad to have flexibility in your machines. Well, we've tried to take that thing that everybody hates and is trying to avoid and say, how can we use flexibility to our advantage, how can we use that to do cool stuff?
[Derek] Now, Professor Howell literally wrote the book on compliant mechanisms.
That's the most cited book in--
[Derek] But he's pretty nonchalant about his work. Just watch how he introduces this mechanism he developed to prevent nuclear weapons from going off accidentally.
Actually the safing and arming of nuclear weapons.
[Derek] What?
And so, yeah, if you want.
[Derek] Hang on, hang on. Hang on, hang on. What in nuclear weapons?
Safing and arming.
Safing and arming.
Yeah. So if there's anything in the world that you want to be safe, that is not going to accidentally go off.
I feel like this is, it doesn't even need saying. But yes, nuclear weapons, obviously you don't want them to go off. Well, I don't understand how this is gonna keep nuclear weapons safe. Now, I wanna come back to this device and explain how it works. Once we understand why compliant mechanisms are best suited to this task. That's cool.
So let's start with something basic.
Probably the first compliant mechanism I ever designed was this thing. What it is, is a compliant mechanism that is a gripper. So you can put something in there and it'll get actually a really high force. I can put that in there and it breaks the chalk.
[Derek] What if you put your finger in there and squeeze it?
You would scream in pain. Would you like to try?
I would.
Okay.
[Derek] Like I would actually like to feel the force.
Okay. You need to squeeze it yourself though.
Really?
Well, all right, I'll squeeze 'til you scream in pain.
But like don't, don't, don't. (Derek screaming in pain) That very quickly got incredibly painful. It felt like having my finger like in a vice.
That looks suspiciously like vice grips, but now with these flexible components where the hinges are.
Hmm. What I learned in my visit with Professor Howell is that compliant mechanisms have a number of advantages over traditional mechanisms.
But I thought he needed kind of a clever, pithy way to remember all of these advantages. So I came up with the eight P's of compliant mechanisms, and the first of those is part count. Compliant mechanisms have reduced part count, because they have these bendy parts instead of having things like hinges, and bearings, and separate springs.
This gripper is just a single piece of plastic, but achieves a similar result to the much more complicated vice grips.
Like how much does it amplify the force?
This will get about 30 to 1. So I could get, for 1 pound force in, get 30 pounds out.
[Derek] That's pretty good. It seems like that would be super cheap.
And really inexpensive. So this we just made here in our shop, but you can imagine also injection molding now.
[Derek] That would cost like cents.
Yep, this would cost cents. The other thing is, because of its shape, you could extrude it and then just chop 'em off.
Hmm.
And that would be cool.
So the simple allows different production processes to be used, which lowers the price. These switches, for example, achieve in one piece of plastic what is normally done with springs, hinges, and many rigid plastic pieces.
Also a good fidget device.
How long can these last?
We've had these in our fatigue testing machine, and we've been able to go over a million cycles without failure.
What do we got there?
All right, Derek, I've got a quiz.
Uh-oh.
A quiz for you, okay?
Elephant.
Very good. Okay, I'm gonna push on the elephant's rump, this direction. I'm gonna hold this. So that little dot right there, is that dot, when I push on it, is it gonna go left, right, up, or down? (Derek laughing)
I just, you know what, I wanted to guess without even thinking about it.
Yeah, please do.
I'm gonna say like up and in.
Okay. Up and in.
And I kind of feel like that because like that would be a logical way for an elephant to hold its trunk. But also because like if this is all going over, I feel like this is gonna kind of extend there and that's gonna get pushed up in there.
Ah, ah, good thinking.
Well, I don't know, is that good thinking?
Well, it's thinking at least.
So this is designed so that when you push on that, it actually just rotates in space, it doesn't move at all.
I knew you were gonna pull some sort of trick question on me.
It's a trick question.
[Derek] Now, since I was fooled by it, I had to try it out on my friend, The Physics Girl.
That's so trippy. That is so cool. I don't understand, what?
It's modeled after the mechanisms you use in wind tunnels where you want to have, say, a model that's attached here, but then you move it and all you want to do is control, its angle and not move it around in the wind tunnel.
Don't displace it, but be able to change the angle.
Devices like this demonstrate that compliant mechanisms are capable of producing very precise motion, which I personally found pretty counterintuitive because these objects are made up of flexible parts. But maybe that shouldn't be surprising, because compliant mechanisms don't suffer from backlash, for one thing.
So backlash occurs when you have a hinge, which is basically just a pin in a hole, and it's moving in one direction. And now, if at some point the motion reverses, it doesn't happen instantaneously because there's some give in the hinge. This also causes wear, and requires lubricant. And that is why compliant mechanisms have better performance than their traditional counterparts.
This one though is my favorite.
That is one of my favorites too. It's just so pleasing, right?
Oh, that sound is so satisfying.
This actually, believe it or not, was inspired when we were doing things at the microscopic level, where we were building compliant mechanisms on chips. We had to be able to make these compliant mechanisms out of silicon, which is as brittle as glass. And if you're trying to make something like this out of glass, it's crazy hard. But that also means, once we figured out the design, we could make it in a material even like PLA, which is also not the ideal compliant mechanism material.
So you can get on our website and get the files to make this yourself.
[Derek] I'll put a link in the description.
Yeah, that also has a nice feel, and a nice snap to it.
It has a really nice snap. I like when it comes out, it's like gunk. Like there's something about that that's really, it's very pleasing. So these things actually move?
Oh yeah, yeah, yeah.
[Derek] I need to see this.
Okay, all right, we'll do it.
[Derek] Were those etched on there?
Yeah, those are etched. And so, just using the same processes used to make computer chips.
[Derek] So another advantage of compliant mechanisms is that they can be made with significantly smaller proportions, because they take advantage of production processes like photolithography.
[Larry] And we have motion that we want at the microscopic level.
That's brilliant. Plus, since they simplify design, compliant mechanisms are much more portable, meaning lightweight, which makes them perfect for space applications.
This here is something we did with NASA, making a hinge that could replace bearings for, say, deploying solar panels. This is titanium, 3D-printed titanium. But what's freaky about it is, you get that motion, which people expect, but here's a piece of titanium that can bend plus-minus 90 degrees, 180 degree deflection.
[Derek] That is solid titanium?
That is one piece of titanium that is 3D printed.
[Derek] There's no alloy, nothing to make it flexible?
Yep, this is, yep. And even freakier than this is this guy right there. So that looks like a crazy beast, but every part in there has a purpose. All these flexible beams. Here are the two inputs. And again, we did this with NASA for a thruster application, where we can put a thruster right there, and now, with our two motor inputs, we can direct that thruster in any direction.
That titanium device moves out, and you notice, it's just all bending. And then, there's no pinch points for the fuel lines, or electrical lines coming in.
Here, this single piece of titanium allows you to use one thruster in place of two.
Okay, that is a clutch.
Okay.
So the idea is, if you spin it up really fast, because it's flexible, this outer part will actually start coming outwards. And then, if there's a drum around it, it'll contact with that drum and spin that thing.
Oh, so this like kind of, oh, that kind of comes out like so.
Then it gets spinning really fast, and then you essentially engage this outer drum. So this is like the way that a chainsaw would work, or something like that, because you get it spinning fast enough and then it engages the chain, and then it turns it over and then, yeah.
The centripetal force. Yeah. Wow, that's cool.
So here, this is made in plastic so that you can see it. But in reality, it's gotta be a lot stiffer. So here it is made in steel.
What? So hang on, you're saying that that thing, which is made of steel,
Yep.
you spin it up to a certain speed, and then it expands and engages a drum that's around it?
Yep, yep. So it'll idle with no motion, but then at a certain speed, what we designed it for, it'll speed up to that RPM.
You speed it up and it engage?
Yep.
I had no idea. Like I have learned something today.
So let's come back to the safing and arming device for nuclear weapons. Its purpose is to ensure that no random vibrations, say, from an earthquake, inadvertently disable safeties and arm the nuclear weapon. Now, one of the requirements was that this device be made as small as possible.
- They made those as small as they possibly could using traditional methods, even using things like what the Swiss watch manufacturers were using.
With compliant mechanisms, they produced a device out of hardened stainless steel where some components were the size of a human hair. This is high speed video. Here, the device is operating at 72 hertz, meaning this little hole makes two complete revolutions each second.
The way it's meant to work is an arming laser shines on the rotor wheel, and when the proper input is given to the system, the wheel rotates a notch. If all the proper inputs are given, then the hole lines up with a laser beam and crazy things happen from there. So it is essential that this device's performance is perfectly predictable, even if it sits unused in a silo for decades.
So are these now being used on nuclear weapons?
You know, it turns out they don't tell us what they do with their nuclear weapons. And so, we designed them, we made prototypes, we tested them, and then it goes what they call behind the fence where it's all classified, and you know, we don't know what happens, so.
But these soft components by themselves aren't truly robots. It's only once you combine them with computers that you get robots which can autonomously form crazy shapes or new styles of movement, all because they bend. But how do they work, and why would you want a soft robot in the first place?
So I came up to Stanford to meet Zach Hammond and his soft robot.
Hey, Derek.
How's it going?
[Derek] All right, you wanna tip it? So is the idea that the robot could walk this way?
Totally, yeah. So you can kinda chain these roles together to kind of roll around in any environment. They call this punctuated rolling locomotion, wherein it's kind of stuck on a face until it tips over, and now it's on a new face, and it can then continue to move its center of gravity. Once that center of gravity exits the support polygon or the base, then it tips over one of the edges of the face.
[Derek] This is a different soft robot made out of flexible tubing. It was designed to mimic the way a turtle walks, where diagonally opposite legs move together. It's powered entirely by compressed air, and perhaps most impressive, it requires no electronics. All of the circuitry is pneumatic. And this means the robot can be used in places like mines, where electronics could spark explosions, or in the strong magnetic fields around MRI machines.
But why would you want a soft robot in the first place?
One of the things that I like to do is just to take the robot and kind of like beat it up a little bit, show how it's compliant and compressive.
[Derek] Well, because they're safer.
If you'd like to take a whack at it, feel free.
But this is your work. I don't wanna break it obviously.
No, feel free. Go for it.
For operation around humans, there's not much damage a soft robot can do to you. I can stand on these?
Yep.
This is a pretty crazy compliant robot.
Because the fundamental structure of this robot is compliant, there's only some maximum force that it could ever exert on me. So it's inherently safe to be operating around people.
[Derek] Could we make it fall and have me be inside it?
Yeah, yeah, we could do that for sure. Just watch your