Robots Searching for Life on Saturn’s Moon with Matt Travers

0:00So Harrison, we had a roboticist in my office from the Carnegie Robotics lab. This is Institute. It was an institute. Yeah, it was amazing. I mean, they might replay this video when the robots take over and go, they could have stopped them then. So this... That was the key point in... This was a part of the timeline where things went dark. Timeline of civilization. Both exciting and terrifying, what can happen in a robotics lab. But what we're trying to do is put robotics in the service of exploration of space. And that's most of what that conversation was about. There could be life on Saturn.

0:36Yeah, yeah, we'll keep looking. All right, coming up on StarTalk. Welcome to StarTalk, your place in the universe where science and pop culture collide. StarTalk begins right now. This is StarTalk. Neil deGrasse Tyson, your personal astrophysicist, I've got with me as my comedic co-host, Harrison Greenbaum. I'm Harrison. Welcome back. Thank you. It's so nice to be back. You were last here, B.C., before COVID. Yes, before the pandemic. Absolutely. Right. Like four or five years ago. Just welcome back. Thank you. And you've been busy since you had a stint in Vegas? I was in the hot, hot deserts.

1:20Whoa. So you've been getting around. It was like a residency in Vegas, right? Yeah, I was the first comedian to headline Cirque du Soleil show. And probably the last. I think they realized they're maybe not prepared for a clown with opinions. Do you have to swing from the rafters? I aggressively campaigned. My first entrance was through a lift in the floor, which is very cool. They give you a safety training because they don't want you to get hurt in the lift. Of course. And this is a true thing. The two main companies that make it are Otis and Schindler. So I was like, let's not have a Jew killed by Schindler's lift.

1:54I don't want to be a part of that obituary. But for my second entrance in the show, I really wanted to come in Miley Cyrus style on a giant bagel. Oh. And I campaigned real hard for it. And I never, that's a dream unrealized. Still to come. Still to come. Exactly. All right. Excellent. So you know what we're going to talk about today? We're going to talk about robotics. Love it. Nobody doesn't love robotics. So we combed the land. Ooh. Yes, we did. And we found Matthew Travers. Matthew, welcome to StarTalk. Thank you very much for having me. Yeah, you came in all the way from Pittsburgh. That's correct.

2:28That would be the home of Carnegie Mellon. That is absolutely correct. And they have one of the greatest robotics labs around. Yeah, that's right. Home of the Robotics Institute. Yeah, so it's not just a lab. It's institutes. The institute. And then the next level is what's society? No, institute, that's a good fundraising level for the title of what it is. So you would call yourself a roboticist. That's correct. Is that correct? Okay. And I've got you here, senior systems scientist. That's correct. Carnegie Mellon. I'm doing well for myself. Okay. Okay, robotics lab, biorobotics lab. That sounds a little scary.

3:05And MATLAB. What is MATLAB? The MATLAB is my lab, for lack of a better description. Oh, MATLAB. The MATLAB. Matthew Travers lab. That's correct. Oh, no, no. No, I will not allow that. Yeah, no, no, it's same for the scientists out there. I don't mean to brag, but we're currently in the Neal Institute. Yeah, yeah. There's another MATLAB, not to be confused with the MATLAB. Okay, I can't believe you went there. Okay. Gosh. But anyhow, we want to just get all into your stuff. Please. So, Matt, tell us, what do you think when someone says robot compared to what your average person on the street thinks?

3:43Because you know robot, it's got to have two arms, a head, and legs, and not sound exactly human, right? It's got to have a robot voice. Yeah. Yeah, for example, there's a gap there. Is one of your goals to close that gap between imagined robots and reality? Or you just don't care? You'll make robots serve whatever the needs are of those who come knocking on your door? Yeah, no, it's an interesting question. So, I mean, I think it's probably philosophical more than anything else. Yeah, well, bring it on. What's a robot? Right, like a coffee maker could be a robot. I would call it a robot. It does a task that you wouldn't otherwise do or want to do or don't care about.

4:22Yeah, it's arguably one of the most, by that definition, one of the most reliable robots at the Robotics Institute. without any question. Works day in, works day out. You know, provides a service. But yeah, I mean, it's a daily activity, right? We work on robots of all different shapes and forms. And it's everything from, you know, sort of more mechanical systems, like you're talking about, things that are biomorphic, things that, you know, look like humans, look like dogs, look like snakes. You know, all the way down to us, what we consider, you know, more on the intelligence side. So what have you been working on lately?

4:55A bunch of different things. So big program with NASA, I think we'll talk a little bit about. I've been working in recycling robots for quite some time, and I've been doing a bunch of work. Wait, wait, wait. Robots that recycle? Or recycling current robots? Yeah, how many people are done with their robots? Yeah, exactly. Now you want to recycle. Well, that coffee maker has been rebuilt so many times, it just wants to end it. No, no. Oh, so these would be robots that know how to sort recycling materials. Yeah, that's correct. Okay, so that's a mechanical thing with some AI built in, right? So there's got to be a whole AI dimension of what you're doing.

5:30Yes. There's a lot of interest in the robotics community and AI right now, sort of like overnight. Right. So let's unpack this a little bit. Obviously, the coffee maker is not making decisions for you. Well, it's grinding the beans, right? No, no, but you told it to grind the beans. Yeah, that's right. It's not otherwise. It's doing tasks for you, but it's not deciding for you. It's not looking at you, get out of bed, and say, we need a triple espresso this morning. But you pop up wide-eyed, and then it gives you a single espresso. So where is the frontier now? Because it seems to me we're capable of building anything mechanical.

6:10Mechanical is just some engineers in a room for a day. It's this decision-making that seems to me the big frontier here. Yeah, related to the question of what actually makes a robot. So for me, when we get down to it, actually combining the mechanical with some form of intelligence and some decision making and some ability to sense the world so that you can get feedback. By the way, we had as a guest on StarTalk, you can dig it up in the archives, the actor who played C-3PO. Awesome. Yes. So at the time, he had written his biography. And in there, what he said was, he's the only person in the world who knows what it's like firsthand to be a robot.

6:54And I said, well, what do you mean by that? And he said, I'd be standing there, and if people chose to ignore me, there's no social cost to that because he's just a robot. Whereas if he were an actual human being, you'd feel some obligation to fold them into a conversation. And so I don't know if this gets to the point of, can robots have feelings? Could you have feelings for a robot? Does that matter? Are there psychologists working in your MATLAB? They're all named MAT. That could guide your morality and your ethics when you invent one of these things? It's become a very big subcategory in robotics has been ethics, roboethics, if you may.

7:43And it's an interesting question. Certainly, you'll talk to people who work with, like, bomb disposal robots. Or people, you're starting to see more, like, dog-like robots, police bomb squads, and, like, we'll have them. And they start to really develop personal relationships. So part of it is probably because people will name the robot. Yeah. And the moment you name it, like, don't name the animal you're going to slaughter for your dinner. because that affects you emotionally. Fair enough. Yeah, I think the emotions that they might be feeling, it becomes part of your team, right? So people do study psychology of teaming and that.

8:19And so starting to extend some of that to now incorporate robots on those teams and the sense of codependency, especially in these jobs that are life critical, for lack of a better description, when you start to depend on something and you have a form of relationship with it. It becomes part of your life as well, yeah. Correct. Wow. Oh, okay. So the future robots should have caution page about - Getting too close. Yeah. And stop there because there's no - My car was named Rebecca. I'll just leave it at that. I won't ask you what you did with Rebecca. So I have a question and this is bothered me is not quite the right word, but it is the right word.

9:01Okay. It is, I've been deeply curious why anyone would build a robot that resembles any life form at all. Just build a robot that is exactly shaped for your task. And that should not have to be something that looks like another life form. Yeah, you're correct. Let's build a bomb-sniffing dog. No, just because dogs have good sense of smell doesn't mean you have to build a robot that looks like a dog. Build a robot that is for the task. Yeah, I mean, people do definitely build platforms that are specialized for tasks, but it comes down to almost more of a philosophical question. Doesn't it have to be true that a specially designed robot will do its task better than a generally designed robot would?

9:52It doesn't have to be. Speaking as an engineer, that kind of has to be the case. Yeah, I mean, what you typically find, one example from stuff that we work on in MATLAB is we design robots that can go up and down stairs. So you can design a robot that's really good at going up and down stairs, but the action is at the beginning and the end of the stairs. So if I design a robot that's only good for that one portion of the task, of course I'm going to have a robot that's really good at going up and down stairs, but going and exploring a floor after it climbs the stairs might be an issue. So there's always going to be a trade-off.

10:36I'm Ali Khan Hemraj, and I support StarTalk on Patreon. This is StarTalk with Neil deGrasse Tyson.

10:51So, how many legs does this robot have that goes up steps? We use quadrupeds. Stairs were made for legs. Trust me when I say that. Legs, but made for four legs, not two legs. Legs in general. I used to walk dogs, and dogs had no— I've seen some footage of past presidents. Having legs is not necessarily all you need to go up and down stairs. Yeah, yeah. So I used to walk dogs for a living, and dogs are completely fluent on stairs, provided they're not bred as city dogs. City dogs don't know stairs. But any other dog, it'll just roll. And they're perfectly fine. And they're pretty good on horizontal ground, too.

11:33I don't think I want a robot dog walker that looks like a dog. Just a robot dog holding other dogs. It's usually all right with it, to be honest with you. No, no, you need the arms to hold the leash. So you have to design a dog with six legs. Oh, God. They can do it. You can do it. Why can't they do it? Hexapods, more legs. Oh, what's it called? Hexapods. Hexapods. That makes sense. Okay. All right. What's the most legs you put on a robot? Six, personally. The way you said it makes it sound like you want to do more. Yeah, he sounds like he's still aspiring. But doesn't six... Octopods. Doesn't six legs maximize a certain efficiency of locomotion because any three points that are not in a line creates a stable foundation?

12:21So a stool is like the minimum thing you can sit on. If it had two legs, you can't sit on it. Okay? You give it a third leg that's not in line with the other two, that's stable. That's the reason you don't sit on a person. Duh. Same two legs. Oh, I see. Okay. So three legs is stable. The minimum stable number of points. If you have six legs, at any given moment, three legs can be raised and moved forward while the other three legs give it stability. those other three legs go down and then it swaps three legs for three legs. Keeping in mind that when we lift up one leg, you're not stable. You have to land once you've lifted it up.

13:06Right. All right, otherwise you'll tip over unless you're a ballet dancer and you don't, I'm not thinking you're a ballet dancer. Wait, what are you trying to say?

13:17So do people think about the stability of what legs are still touching the ground versus which others can move forward? I think that's how beetles can scurry so efficiently. Their legs do this thing. Yeah, so it's called an alternating tripod gait, as you might imagine. We got a term for it. Yeah. Okay. Alternating tripod gait? That's correct. Didn't know that. Thank you. You're welcome. I'll say next time instead of spending five minutes explaining it. Okay. I was going to cut in, but I didn't want to be rude. Oh, no! Be rude. Be rude. I was going to call it the mat gate, but it might be too much.

13:49But you're correct. It might not be the most efficient, may be the most effective in terms of maintaining static stability. So with humans, right, sort of the main difference between what you're talking about with hexapods and humans is we actually have a dynamic stability, which is sort of a different... The way a bicycle has a dynamic stability. That's exactly correct. Right. If you don't pump energy into the mode of going forward, you're going to fall over. Right. Interesting. But on the flip side, as I think a famous robotist once said, mo' legs, mo' problems. Yeah, I think it might have been me.

14:25Is that a problem where like if you have too many legs? I feel like you have to know where each leg is going to be at all times. The middle P didn't seem to have an issue with this, okay? It's an interesting question. So when we start to get into robots with even more degrees of freedom, for each of those legs as a roboticist. So tell me about degrees of freedom. Yeah, sorry, excuse me. So degree of freedom is like in your elbow joint or any of your joints is a degree of freedom. Typically one joint, so like a simple hinge joint, you can get one degree of freedom. Got it. But the rotation, I guess that's also, so now I'm moving but not moving to a new place.

15:01Yep. So is rotation a degree of freedom to you? Yes. Yeah, okay, good. Rotational and translational. Translational, all right, good. Yeah, so as you start to add more and more degrees of freedom, you need to start to be a little bit sophisticated in how you are going to coordinate this, right? I mean, I have lots of degrees of freedom in my body, but I'm able to coherently put those into forward motion to walk, to jump, climb stairs. You're not a badly connected skeleton trying to walk down the street. Yeah, that's correct.

15:31No skeleton has ever had swagger. They're always ready to crumble into a pile of bones. So as you boost the degrees of freedom, it comes with higher responsibility. It requires more control. Yeah, you can articulate more. So the overall number of, I'll lack of a better description, degrees of freedom that you have, and the expressiveness that you can get out of more degrees of freedom, the number of places that I can weave into and, you know, whatever, ad capability does go up, but it comes at the cost that it becomes more complicated to coordinate all of those degrees of freedom. For me... Into some coherent project or task.

16:10That's correct. I got into a little dust-up on The Late Show, with Stephen Colbert because we had both seen Dune. We went to the world premiere and I didn't hang out with him there, but we saw each other and I was on his show shortly after that. So what'd you think of Dune? Because he loves these big fantasy stories, right? He's a big fan of Lord of the Rings. So what'd you think? And I said, well, the worm, okay. But there's a scene where the worm moves really fast and it is straight as it does that. The worm has no legs or arms or wings, so you can't take a straight thing and just zoom down the highway, okay?

16:55If you ever see a snake, it has to coil and then propel itself off of that. If you ever see a stretched out snake, it is harmless to you, okay? It can't just all of a sudden bite you or chase you down. That's not gonna happen, okay? Straight things are slow. That's why the gay pride parade is so much quicker.

17:14So fast. Nailed it. So what good would a robot be that has no arms or legs? Because I'm reading up that you're collaborating with NASA on this EEL project. Okay, this sounds like a tortured acronym. Let's see what they pulled out of this. Exobiology Extant Life Surveyor. Did I get that right? Eels. Eels. Okay, so this is, it's a robot. It's a robot. In design right now, presumably, right? When will it be ready and deployed? I believe 2028 is target. In just a few years to come. That's correct. And what is its prime directive? The scientists at NASA believe that the greatest probability to find other biological life in our solar system is on one of the moons of Saturn, Enceladus.

18:13Okay. It's the name of the moon. It's a ice-covered planet, and you can see, so they've sent previous missions there. In this context, he's allowed to call the moon a planet. Okay. Because it's a world, okay? It's a spherical world, and I'm giving him that, okay? But don't come at me and say, but how about Pluto? No, man, no. Okay? Okay, go. My bad. All right, so previous missions have found these geysers, There's plumes of whatever material coming up from the surface. Coming from below the surface through the ice. Correct. Because there's pressure down there, and any time pressure and some fluid, something's got to give.

18:49I'm keeping it so clean. I'm keeping it so clean right now. Thank you. You shackled to the chair.

18:59So they believe, yeah, that pressure is being created by subsurface ocean. And they believe that, so they, the scientists at NASA, believe that that has the highest probability for finding life. Now, if you have geysers, you don't have to go to the ocean because it's coming to you. How does that help or hinder your efforts? Yeah, that's one of the theories that you could fly through the plumes as they're happening to try to collect data there. There's other thoughts that you can just go to the surface and start taking samples there. Because the geyser comes back down and it's... It's water? Yeah.

19:30Well, excuse me, we don't know. It's very likely water. Very likely, yeah. So there's a couple different theories on what would be the best way to go search for life there, which is ultimately what they're trying to do. With this, with the Eels Project, we're actually trying to get into the ocean via the sinkholes. Oh. Do we know? Have you considered knocking?

19:52So that means we have some idea how big the opening is. I didn't know that. That's interesting. Sort of. I believe the resolution of surface, any pictures that we have of the surface of Enceladus are about six meters per pixel. So I believe that the openings into whatever, the sinkholes, we think they can see them. Okay. But not with great enough resolution that I think you could measure down to meters. All right. So it's got to be worm-like to get in the hole. Yeah. And then what does it do? So this project, basically, we're trying to, in the same way, like a boa constrictor, actually exactly the same way can constrict around something with snake robots, which is the eels robot is basically a snake robot.

20:37It doesn't look like an eel? Good one! Wait. He needs a new napkin. Wait a second. The eel robot is designed to look like a snake? Eels and snakes look alike. They look alike, but they're not the same. Harrison uncovered the flaw in your project. Okay? This is unbelievable. Snake-like robot tried to climb down the geyser by pushing out against the sides of it. We're not going to drop a robot basically down the geyser, so we want to do a controlled descent by pushing out. Yeah, this is like very Mission Impossible kind of. Very much. You know, yeah, okay, cool. Absolutely. So you have things that protrude from the side of your eel.

21:14It is using its body, the body of the robot, to actually push out and create friction against the sides of the... So does it do it in a curvy sort of way? So it can span whatever the width is just by curving to do that. Yeah. Interesting. Yep. Now you designed it to get in, but now you want to do interesting things. How's the eel going to do that? So that's where I think the eel comes in, is it actually, we want to be able to get down through a hole and then swim in ocean and take samples and search for life. So where are the sensors? So right now the sensors are in the head and in all of the individual degrees of freedom.

21:50So the degrees of freedom in this case all have motors attached to them as well. So you can control them. Okay. And does it have to come back out? Excellent question. So as of yet, once you collect the information, which you have to do physically, collect the samples, getting the information back out, you can do wirelessly. Right? If you have something flying over, you have line of sight. Got it, got it. Okay. So whatever it's doing, it's information in situ, and then it sends that information back about whatever it collected. One way to do it. So as of yet undecided, I think, for that program.

22:23Okay. So you're leaving that poor eel? And we're talking about personalizing. Don't name this thing. Don't name it. You're going to feel real bad abandoning it on Saturdays. I don't think anybody has feelings for eels. I'm sorry about that. Is it furry? Doesn't know? Doesn't have big eyes? So this is not designed for a sample return then. It's going to get all the samples it can. Yeah, I don't think the answer to that question yet exists. So you made quick mention of the constricting part of a boa constrictor. In what way is your eel constricting? So it's doing the opposite or like the complement of constricting.

22:57It's actually pushing out. Oh, okay. So it's the same dynamic, I guess. So you're using forces. Instead of like a normal robot you think of in a factory where it's doing pick and place type actions, you know, very robotic and mechanical. In this case, it's more biological in the sense that it's trying to engage with the environment. It's trying to actually feel its way proprioceptively. In this case, feel its way through the environment. So Saturn is an hour and a half away light travel time. So no one is giving it fast commands. Look out for the monster. It's too late. The subsurface monster.

23:35So it's got to have some decision-making abilities. What might that include? 100%. So yeah. He's just looking outside the hall. I'm not going in there. Get out of your mind. That looks dangerous. You go first. Yeah. This is so cool. So we just had actually a paper in Science Robotics that just very recently came out that actually— Wait, wait. That's a journal. That's a journal. Science Robotics. That's correct. Yep. Wow. So that article goes into in great detail exactly what the autonomy system is. But in general, you know, very, very high level. You have sensors. You have your platform, which can move around, interact with the environment.