SciShow Tangents - Robots

Starting point is 00:00:00 Hello and welcome to SciShow Tangents, the lightly competitive knowledge showcase starring some of the geniuses that make the YouTube series SciShow happen. And this week, as always, I'm joined by Stefanfan chen i've joined you what's your favorite season summer is it one the one that california is all the time just as much hot as it can be i'm so hot and dry what's your tagline all them canned goods oh all of them that's That's the way to go right now. Sam Schultz is here as well. What's the best grid pattern for planting plants in Animal Crossing New Horizons? Oh, well, you know what, Sari?

Starting point is 00:00:52 Just forwarded me a very interesting link. Five by five. It's more complicated than that. I won't get into it, but. Okay. And what's your tagline? Well, it was going to be five by five flower grid. You kind of ruined it for me.

Starting point is 00:01:05 Oh, sorry. Sari Riley is here as well. Sari, who is the Tiger King? Oh, I don't know. He's like a man with a bleach blonde hair and mustache. I have avoided watching the Tiger King. It's probably for the best. Yeah, anything that I know about him is just myth and legend.

Starting point is 00:01:23 And it seems like the show is so wild that you could tell me anything and I would be like, sure, that happened. Like a man played a kazoo and rode a tiger around. I'd be like, yes, that's the premise of The Tiger King. Sari, what's your tagline? Marshmallow Surprise. And I'm Hank Green and my tagline is 10,000 wipes. Every week here on SciShow Tangents, we get together to try to one-up, amaze, and delight each other with science facts. We're playing for glory, and we're also keeping score and awarding sandbox.

Starting point is 00:01:50 From week to week, we do everything we can to stay on topic. But we're not great at that. So if you go on a tangent and the rest of the team deems that tangent unworthy, we'll force you to give up one of your sandbox. So tangent with care! Now, as always, we introduce this week's topic with a traditional science poem this week from Sari. Flashing lights, whirring motors, two legs made of steel. An unfathomable head with sensors concealed. What powerful feats this metal beast might complete.

Starting point is 00:02:16 Oh, it just did a backflip. That's freaking sweet. We imagine our bodies, our movements, our thoughts wrapped up in circuitry and made into mascots. But that concept, dear humans, is inherently fraught because robots are not exactly what we were taught. Snake-like tubes or big welding arms, watering farms or sounding alarms, little vacuums that clean while doing no harm. Even soft robotic prosthetics have their own charm. Whether a machine is uncanny or a chunky space probe, the size of an elephant or a tiny microbe, one thing rings true in every scientist's heart, we'll never stop programming robots to fart.

Starting point is 00:02:53 Who's the farting robot? I didn't know about this either. There is a thing called the robot. It is an interactive farting figurine by WowWee. It's available at GameStop and Walmart. This is really interesting because I also found a robot, same name, but it's used by Ford to test out car seats. And so it sits like a human male would in a car seat.

Starting point is 00:03:20 And it's sweaty and just like a butt that sits on car seats. Does it communicate its own comfort to to you no i think it like is to simulate wear and tear on the car seat over years because people are trading in their cars for new cars less often right it's about the impact on the seat not not the impact on the butt yes i'm willing to take that job and sit on a seat 100,000 times. Yeah, this is why they gave that job to a robot, Stefan. They can't afford you. Sari, what is a butt?

Starting point is 00:03:53 I mean, what is a robot? Well, a robot like a butt has an impact on the physical world around us. And so a computer program that stays contained within a machine and does calculations, that's not a robot. But if it can interact with the physical world, I think that's What about like a CD player then? Because it like spins a CD. I guess that's a kind of a robot. Yeah, I guess kind of. It seems like scientists don't have a distinct definition of a robot. The Robot Institute of America says it's a reprogrammable, multifunctional manipulator designed to move material, parts, tools, or specialized devices through various programmed motions for the performance of a variety of tasks. So, like, a CD player fits into that, maybe.

Starting point is 00:04:43 Well, what's the difference between a robot and a machine? I think robot starts to get at, like we've been talking about doing tasks. So like a machine is like a toaster that you have to manipulate yourself. So like you have to be the human finger to push down the toaster button. But a robot toaster would like grab the bread for you

Starting point is 00:05:04 and stick it in and then go and then pop it back up and you would have to do nothing like it would replace you so you wouldn't be necessary in the toast making process anymore okay i feel like we sort of had the idea of what a robot was before we had robots so it's like well a robot is a is a machine that does human things and looks human and acts human but like can can do more than a human can physically at least like making and then we were like yeah and then we were like but actually we're gonna have real robots but they're not gonna fulfill a lot of those categories because most robots do not look anything like people um because their goal isn't to replace people. It is to do an action.

Starting point is 00:05:46 And so it is just the one part of a person that is most useful to it, like an arm. Yeah, so all the science fiction robots that we see are very humanoid, usually programmed with some sort of artificial intelligence so that they can make decisions about a wide range of things. But that technology is very, very far off and doesn't exist. But a lot of our robots are just like, I've programmed this

Starting point is 00:06:12 thing to sit on car seats. I mean, the Robot Institute of America or whatever the organization was said it was a programmable thing, which was interesting to me because a lot of what we think of now is like robots that kind of program themselves. And in the same way we can make decisions that can use sort of like simple artificial intelligence to figure out its own decision making or its own object identification. So it in some way is doing its own programming. So it's almost like if we have that kind of artificial intelligence or even like a generalized artificial intelligence, it stops being a robot at that point and becomes something else. Sari, what is the etymology of the word robot?

Starting point is 00:06:52 Because I know it's interesting, but I've forgotten. So the word robot comes from a Slavic language, I think Czech or whatever they were speaking in the early 1900s, called robata, which means forced labor. And it was coined in a play called Rossum's Universal Robots by a playwright called Karl Čapek. And it was just about mechanical men that are built to work in factory assembly lines

Starting point is 00:07:22 and that rebel against their human masters, which is like classic. Robots are always doing that. Yeah, robot uprising story before robots were a trope. Yeah, I mean, it was, yes, a trope has to start somewhere. And then Isaac Asimov used it and the word robotics in a short story. And he was a little bit more optimistic about how robots would help out humans instead of be part of an uprising. And he was the one who came up with the laws of robotics that robots won't harm humans and things like that. Yeah, the great thing about the laws of robotics are that they really require a huge amount of understanding on the part of the robot, which we are nowhere near.

Starting point is 00:08:01 It's like, how do you know when you've harmed a human? Seems very obvious to me. I know when I'm harming a human, but boy, a robot could downright destroy you and have no idea it did it. Now it's time for Truth or Fail. One of our panelists has prepared three science facts for our education and enjoyment, but only one of those facts is real and the rest of us have to figure out either by deduction or a wild guess, which the true fact if we get it right we get a sandbuck if we're tricked then stefan will get the sandbuck stefan what are your three facts okay these are three facts relating to water-based robotics oh uh so fact number one a uk-based hobby roboticist created a bunch of robotic versions of different Some researchers who heard about this got interested and were particularly interested in his plesiosaur robot

Starting point is 00:08:58 because the plesiosaur is a little bit unique amongst animals because it has four identical flippers. And so there's been some mystery about how it moves. And so by studying his robot, they figured out how plesiosaurs move their flippers in relation to one another. That's fact number one. Number two, the modern conditions around the Great Barrier Reef have led to a population explosion in giant reef-eating sea stars, which has led to the development of a fleet of autonomous robots that can patrol the reefs, looking for these sea stars and delivering lethal injections of bile salts to kill them. What salts? Bile salts.

Starting point is 00:09:35 Bile, like the stuff that your liver, or gallbladder. Gallbladder. Yeah. Or fact number three. One company is taking lifeboats to the next level by turning them into autonomous firefighting watercraft that once deployed can automatically navigate to humans that are stranded in water as well as fire water cannons at flames that it spots on the vessel. So we've got fact number one, a hobby roboticist decided to race some swimming dinosaurs and this taught us potentially how real plesiosaurs swam. Number two, there are some bad sea stars

Starting point is 00:10:08 on the Great Barrier Reef and that's led to the development of a fleet of autonomous robots that can look for and inject lethally the sea stars with bile salts. Or three, a company that is taking lifeboats to the next level by turning them into autonomous firefighting watercraft that can automatically navigate to humans stranded in the water and shoot water cannons at the flames.

Starting point is 00:10:33 Not fire cannons. That would not be helpful. Shoot fire cannons. I feel like I have heard about people making like racing robots for fun and also for science. So this one has credibility. This first one has credibility for me. That like one of the ways, and I have also heard about lots of like swimming robots and how we're going to figure that out.

Starting point is 00:10:58 But I really liked the idea that like somebody was just having a fun. idea that like somebody was just having a fun and then they were like actually can you send us your video because we're a little bit confused about how plesiosaurs work yeah i could totally imagine a scientist nerd like looking on a shelf and being like oh i have to see these dinosaurs wonder if i can make a move and then doing that because that seems like what i would do if i was bored and had electrical engineering skills. Yeah. Like I don't think that I could do it

Starting point is 00:11:27 but I could imagine how someone might make a robotic plesiosaur fairly easily. They're the ones with like the paddle-y kind of. Yeah.

Starting point is 00:11:35 They got like paddle fins and they got a long tail and a long neck. Sounds like the plot of like a Mega Man game or something to me. Doesn't sound real.

Starting point is 00:11:44 The Seastar one sounds too sad and also a little bit too specific me. Doesn't sound real. The sea star one sounds too sad and also a little bit too specific. Are there giant reef eating sea stars? Is that a thing? I have no idea. I know that sea stars eat all kinds of stuff, but I do not know about the sea star situation on the Great Barrier Reef. Yeah. And I don't know if sea stars would have any need for eating eating bleached coral or if they were eating whatever is alive. I'm sure they'd be eating the living stuff.

Starting point is 00:12:10 Like whatever's left. Yeah, so I guess I can see a case to protect whatever's left then if they were these giant. Do you know, Stefan, if these sea stars belong there? Are they invasive? How'd they get there? I'm not 100% sure, but I think they do belong there.

Starting point is 00:12:26 So let's ask more specific questions about other things then. I think I talked about some kind of algorithm that can take a census of what fish live in a reef. So maybe it's some kind of adaptation of that idea. And then, I mean, an autonomous lifeboat that can rescue people and shoot water. Like that feels real because like if it's not being done just upon hearing it, I'm like, that's not a thing. I should found an autonomous lifeboat company.

Starting point is 00:12:52 Machine learning could easily know what fire looks like. That's very easy. And one of the great things about autonomous boating, like there's just less stuff to run into in the ocean. On roads, like it's very easy to like leave the road and that's a big problem. On the ocean, it's very hard to leave the ocean. All right, who's going to guess first? I'm going to guess dinosaurs just because I think it's fun.

Starting point is 00:13:20 It is fun. Sam, hit me. I might go with dinosaurs too because I think boat seems slightly too boring to be the right answer to me. That's true. It is a little bit boring. And I'm going to go with starfish because I know that that is the correct answer

Starting point is 00:13:32 because I've read about this. No! Hank is correct. It's the starfish. So there's, apparently there's the three major threats to the Great barrier reef are climate change pollution and these sea stars yeah they're nasty yeah so they're called the crown of

Starting point is 00:13:53 thorns starfish and they're one of one of the largest sea stars and they're about a foot wide um and they're covered in these venomous spines um and over the past decade or so, their populations have boomed a lot because all the agricultural runoff going into that area causes these algal blooms and the sea star larvae are eating that algae. So they are having a grand old time over there. And we also, they had some natural predators, but we ended up overfishing those predators. So it doesn't have any like checks to its population. robots except robots so they do eat the corals um once they reach like maturity they eat the fast growing corals i guess which is good if you have a little bit of that because it makes some room for the slow growing corals

Starting point is 00:14:38 to establish themselves but it's i guess it's estimated that these sea stars are responsible for about 40 percent of the overall coral loss that we've seen. So it's a pretty big deal. Mr. Matthew Babadin, who's a professor at Queensland University of Technology, was all the way back in 2005 starting to develop these systems. And they had like sort of rudimentary vision system that could recognize the sea stars like two thirds of the time. But at that time, they didn't have a good way to kill them. They had like some kind of lethal injection, but you had to inject all of the 20 arms.

Starting point is 00:15:09 And so it was like really difficult to do reliably. But in 20- And this is something that like people would do. Like divers would go down, stab 20 different legs of a starfish and then move on to the next one. Got one. And it's venomous too, right?

Starting point is 00:15:22 So they were trying not to get stabbed back. Yeah, they have these like long long stabby poles uh but by 2014 we had found this bile salt injection thing that uh has a 100 mortality and you only have to poke it one time uh and then by that point his uh vision system was capable of identifying the sea stars over 99% of the time. They went through a couple iterations, I think, but they ended up with what they're calling the Ranger Bot. It's about a meter long. It's got a bunch of propellers so that it's really maneuverable.

Starting point is 00:15:55 And I guess the battery lasts about eight hours and it can go at night. It can go during the day. If there's a storm out, it can go anytime. And it's super easy to control. I guess they did a lot of user testing to make sure that it was user friendly so you know people who are trying to save the reefs can go out and plot courses for these robots and control them and as a bonus it has a bunch of like sensors on it so it can monitor the reef health while it's out there looking for these sea stars and killing

Starting point is 00:16:21 them uh and last i had read, there were about, I think there were five of them that were operational, but they're not yet widely available to different like reef management teams. But the idea is to have a bunch of these like fleets just all over the reef. That's cool. Next up, we're going to take a break

Starting point is 00:16:41 and then it'll be time for the fact off welcome everybody sam buck totals sari has one i have one stefan has two and sam has none sorry i put you in the end there i don't know exactly why i did it in that order but now it's your chance sam because it's time for the fact where two panelists have brought science facts in an attempt to blow everyone else's minds the presentees each each have a sandbuck to award to the fact that they liked the most. And we will decide who goes first with this trivia question. As Sari said, the word robot first appeared

Starting point is 00:17:33 in Czech playwright Karel Capek's play, Rossum's Universal Robots. In what year did this play premiere? I'm going to say 1917. I'm going to say 1904. Okay. Hank wins. It was 1921.

Starting point is 00:17:59 Oh, nice. Okay, we were pretty close. I want to go first. So once upon a time in the old days, rich people had people put their clothes on for them. But that is still something that happens for some people who need help getting their garments on and off. And there's plenty of reasons why, age, injury, other kinds of limitations. So scientists at the Georgia Institute of Technology have gotten a robot to start to figure out how to dress a person. So they used a pre-built research robot. This is a thing that already existed called the PR2. And it can be programmed to do things like fold towels or grab drinks for people from the fridge. They wanted to learn how to dress a person which means you have to let it fail and make mistakes but robots making mistakes with a real human body would be dangerous because as we discussed earlier

Starting point is 00:18:54 robots do not know when they are killing you so they had a robot study 11 000 simulations of a robot putting a hospital gown onto a human arm. And it had the robot analyze the kinds of forces it can apply and the motions it can make and how those forces and motions affect the person who is getting dressed. In some of those simulations, they intentionally had it go very wrong. So like the gown would catch on the hand or the thumb or the elbow and to deal with it the simulated robot would then apply a dangerous level of force to the arm and those were given to it as intentional failure states so it would know this is bad this simulation went very wrong never ever do this so it went through 11 000 of these simulations and it got through

Starting point is 00:19:46 them in one day because it's a computer and it can do that and then it moved on after that day to dressing people and it was able to do that by which i mean it was able to put one sleeve on one arm of one person uh in about 10 seconds which is like, you know, maybe not as fast as I would do it, but plenty good. Importantly here, it is using touch so it can feel how things feel on its fingertips to figure out what the person getting dressed might be feeling. And also it's using its sight. So it's feeling and watching and using all of that information at the same time. And then from all of those movements and all the simulations and all the data it's getting it can sort of pick the best motion for getting

Starting point is 00:20:30 the the arm into the sleeve which so far so good uh as of 2018 it was able to put the surgical gown on the arm of a person getting a person fully dressed will take more work, but we're on our way. Yeah, it's worth it if I never have to put my own pair of pants on ever again. It's also worth it if the robot doesn't rip your thumb off when it's trying to put your shirt on. All right, Sam, what you got for us? All right, so one big hurdle encountered when making robots that are intended to interact with and walk around in their environments is maneuverability. interact with and like walk around in their environments is maneuverability. So when people and animals move around, they're balancing, they're like pathfinding, they're adjusting to changes in incline and they're like jumping around and they're even transitioning from like water

Starting point is 00:21:14 to land to sea. And not to mention that robots run on batteries and they can't just stop and eat like a bug or a bunch of grass when they need to keep going. Not yet, at least. And there have been a lot of advances in robo-mobility, but they can't really compare to good old-fashioned flesh and bone. So researchers at the Korea Advanced Institute of Science and Technology took a kind of weird and freaky shortcut. They developed what they call a parasitic robot system that commandeers an organic being and pretty much uses them as like a horse so their first and i think the only test subject that they've done so far were a bunch of turtles

Starting point is 00:21:54 red-eared slider turtles uh they were chosen not only because they're amphibious so there would be lots of options for the different kind of terrain they could do but they have good memories and they come with a big old shell that you can glue a bunch of electronic components to so the robot is basically like a little microchip brain hooked up to a battery and then a bank of five red led lights that are mounted horizontally in front of the turtle's face and then a like a little container of food that's like a gel and a spray nozzle that they position near the turtle's mouth. So for two weeks before they started this experiment, the turtles were fed while they were looking at a red LED light.

Starting point is 00:22:33 And then they put the robot on them. And the robot had instructions to move the turtles along certain paths. So to do this, it would light up one of the five red lights in the direction closest to the way they wanted the turtle to go. And if the turtle went the way that they wanted it to, the turtle would get a little gel treat from the robot's food tanks. So after five weeks of doing this, the robots were guiding the turtles through 16 feet of track in 75 seconds with a deviation from the ideal path of less than 3%. So one of the big challenges

Starting point is 00:23:09 that the researchers faced was that the turtles would sometimes get distracted by stuff that wasn't the red lights in front of them. So future experiments they are planning will use full virtual reality turtle headsets to ensure that the parasitic robots have complete and total control of the

Starting point is 00:23:28 turtles. Oh my gosh. This is way less bad than I thought it was going to be. It's still scary though. Yeah, I thought we were going to be drilling holes in these poor boys' heads. I don't know. I think they just glued the things onto their shell. I don't think any holes were drilled anywhere

Starting point is 00:23:43 in the turtles. I'm very impressed that turtles are this trainable. I had never really thought about trying to train a turtle. Yeah. I sense that I am in trouble. Do you guys want to choose between the two facts? We have 10,000 simulations, some of which were violently incorrect, leading to a robot that can put a sleeve onto a person in 10 seconds.

Starting point is 00:24:04 Or Sam's amazing parasitic robot turtle. Three, two, one. Sam. Sam. Yeah, I know. I think only because the dressing robot only got one arm. If it had gotten the other arm, I would have been like, yes. But they had complete control over these turtles.

Starting point is 00:24:26 That's right. They did. Unless the turtle got distracted. Yeah. Hank, you did too good at science journalism where you were like, okay, I'm going to lower your expectations one arm instead of starting with a fully automated closet that addresses you is on the way and that means it's time for ask the science couch where we have some listener questions for

Starting point is 00:24:54 our couch of finally honed scientific minds at treehouse down asks why is robotic skin so hard to make well it depends on what you mean. So like just covering something in plastic is not hard, but if you want it to sense, that is very hard, it turns out. So there are like a number of reasons why. One, because like we sense many different things. And two, because like the nerve density of our ability to sense and then and then to send that information uh for each little bit of skin like it's it's amazing that we can do this but if you're trying to do it with a

Starting point is 00:25:32 robot you have to have each like tiny bit of resolution of skin feeling resolution to like have a separate wire that connects to a freaking central processing unit and that is just it's it's miserably difficult so that's one of my understandings at least of this who is making robot skin and why are they making robot skin do you really have to ask not just sex if you want any robot to be able to, so like take the arm robot that has to touch someone's arm and be like, oh, I'm going to put a sleeve on this. It needs like touch receptors to know how much force it's putting on that arm. It needs probably like temperature sensors to be like, is this a living human or a dead corpse? I don't know.

Starting point is 00:26:24 That's like kind of a bad example. That's important. You want it to understand the world around itself. So if it runs into something, you want it to know that that happened. And a lot of times right now, it just literally can't know. And if it does know, it knows that something happened, but it doesn't have any idea what it ran into or in what direction. Yeah, and the way that we have spatial orientation, It doesn't have any idea what it ran into or in what direction. Okay. Yeah.

Starting point is 00:26:45 And the way that we have like spatial orientation, we know how our body is arranged relative to itself. So we like know our arm is to the right or to the left of our body, for example. That's all nerve endings. It's called proprioception and it's in our muscles and our skin. and it's in our muscles and our skin. And so like in order to have, especially humanoid robots, but any robot that's doing a delicate task,

Starting point is 00:27:09 you need some equivalent of skin with all these sensations to do the delicate actions. But as far as answering the question, Hank is right. It's mostly just because our skin is so dang complicated. In addition to all the wiring stuff, our skin can get damaged and still function. And that's a hard part of approximating skin. So even if we have a stuff, our skin can get damaged and still function. And that's a hard part of approximating skin. So even if we have a cut in our skin, that doesn't mean all the nerve endings are suddenly destroyed. But if you have a cut in robotic skin that slashes through a sensor

Starting point is 00:27:35 that could mess up the whole system. Right. And so a lot of innovations in robotic skin technology are electrical engineering related and have to do with programming the electronics and the signals so that the processor that is receiving all of them doesn't get overloaded with information. There are a couple different ways that people are experimenting with it. One is called asynchronously coded electronic skin or ACEs, which I think it's similar to this other one. The way it sends signals is not all at the same time. It spaces

Starting point is 00:28:11 them out in such a way that there isn't like a big backlog of signals like waiting to be processed. And then another one, it's like above a certain threshold of activity. So like if you put a hat on your head, your head senses it. It's like, oh, that's weird. There's a hat on my head. But after a little bit of time, your head just becomes used to it unless something else changes, like you take it off. And so they're trying to program a robotic skin to mimic that. So like it recognizes the change in temperature or pressure and then recognizes it for a time. But then when it becomes part of the robot state state of being you ignore it so that you can focus your processing energy on other things we kind of do that too we're like you start to tune things out after the signal has been there for a while yeah

Starting point is 00:28:56 it's like how you don't know where your tongue is until i said that now you're like i always know where my tongue is sam's constantly thinking about his tongue if you want to ask the science couch your questions follow us on twitter at scishow tangents where we will tweet out the topics for upcoming episodes every week thank you to at camilla 13 and at little chris and everybody else who tweeted us your questions this episode sam buck final scores sary you and I are tied for last with one point. Sam and Stefan are tied for first with two. So for the season, that brings us to Sari in the lead with 37 points,

Starting point is 00:29:32 followed by 36, Stefan, 35, Sam, and 34, Hank. Oh, no. It's tightly packed. If you like this show and you want to help us out, it's real easy to do that. First, you can leave us a review wherever you listen. That's helpful and helps us know what you like about the show. Second, you can tweet out your favorite moment from the episode or just say nice stuff to us on Twitter.

Starting point is 00:29:51 And finally, if you want to show your love for SciShow Tangents, just tell people about us. Thank you for joining us. I've been Hank Green. I've been Sari Reilly. I've been Stefan Chin. And I've been Sam Schultz. SciShow Tangents is a co-production of Complexly

Starting point is 00:30:04 and the wonderful team at WNYC Studios. It's created by all of us and produced by Caitlin Hoffmeister and Sam Schultz, who is also our editor. Our editorial assistant is Deboki Chakravarti. Our sound design is by Joseph Tuna-Medish. Our beautiful logo is by Hiroko Matsushima. And we couldn't make any of this without our patrons on Patreon. Thank you. And remember, the mind is not a vessel to be filled,

Starting point is 00:30:26 but a fire to be lighted. But one more thing. If you were a medical student in the UK in 2016 and you needed to practice performing rectal exams, your pickings were pretty slim. In fact, there was only one person in the whole country signed up to allow med students to perform practice exams on them, which seems like kind of a problem. I mean, it seems like it would either be zero or more than one. Just one. So a team at Imperial College London got to work inventing a robot ass. The result was what looked like a disembodied butt filled with little pistons and robot arms surrounding a silicon tube that was like a rectum. And they could could the pistons and arms would squeeze to provide

Starting point is 00:31:25 different amounts of pressure to simulate rectums of all different shapes and sizes and it can also simulate different diseases and complications of the prostate so you just have a robot butt you can dig around in now

SciShow Tangents - Robots

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.