SHIFT_E98_Space Robots_SEG C
[00:00:00] What's an astronaut's time worth? Answering that question is not a perfect science, but for someone on the International Space Station, it's probably somewhere north of $100,000 an hour. That's when you take into account what it costs to keep them alive up there, let alone their commute, plus a salary.
Whatever the exact number may be, it's safe to say it's on the high side of what you wanna pay for bagging up trash and unpacking groceries and other supplies. This week, we take a field trip in New York City to see some space robots that could help with this problem. I'm Jennifer Strong and this is Shift.
Now, here on the steering column is a device called Autocruise. You simply set the speed you want- Self-driving robo-taxis are already on the road in China. US cities. As the disc rotates, a mirror reflects the light in the way that depends on how the signal was recorded. This[00:01:00] 
is the 100 terabyte hard drive. I present to you Electro, the mono man. Ladies and gentlemen. I would say that one of my greatest skills is my ability to interact with humans. This episode, we head to the Brooklyn Navy Yard to visit the home of Icarus Robotics.
You gonna show me some robots? Yeah. Yeah. That's the fun part. So we like to make a joke that the company started with just two guys in a room with, you know, some cheap robot arms and a cargo bag from NASA. Hi, I'm Ethan Barajas. I'm the co-founder and CEO of Icarus Robotics. And so we were lucky enough to schmooze one of my old advisors from [00:02:00] NASA out of a space flight-ready cargo bag.
So this bag that you see here on the table has been to the ISS. It's brought cargo up. Everything that goes to space goes to space in these bags. And these crews of, you know, four astronauts at a time will take up to seven days unpacking this visiting vehicle of three and a half tons of these same, all looking similar white bags, and then repack it with down mass.
So this is all the waste, all the trash, and that takes another seven days. And we're doing this every 60 days. So to burn two entire weeks on just moving bags, you really lose out on all the science and manufacturing. I started working at NASA when I was 17 on the autonomous growth of plants for the ISS, and that was my first real internship.
And I got to see that project through all the way to its end and work on that for a few years and actually send it to the space station, which was super exciting. And then from there, I went to Caltech and I got to pursue my degree in mechanical engineering and work with some really amazing roboticists, some amazing professors at JPL, and some amazing lunar rover and Martian rover projects.[00:03:00] 
And so I got to work on this one really amazing one with Clint Carpenter and Professor Soon-Jo Chung on, uh, Chariot, and it was next-generation lunar rovers for what's called PSRs. These are permanent shaded areas or regions, and these are just craters on the Moon to find water ice. And, uh, then I was tapped for an accelerator called Entrepreneur First, and that's when I was introduced to Jamie Palmer, my co-founder and CTO of Icarus.
Hey, so, uh, it's Jamie Palmer here, co-founder and CTO of Icarus Robotics. So yeah, I guess s- some of my story was, did mechanical engineering at undergrad and was really interested in motor sports. So I pursued that really far and ended up doing Formula Student, building small-scale race cars, bringing them to tracks, uh, across the UK and Ireland.
And then eventually I ended up, uh, landing a job in F1, so I spent a year at the Mercedes F1 team, uh, doing test engineering. But during my time as well, uh, before that, I got really into robotics on the side, so doing research, uh, just in my spare time for no credits with, uh, one of my professors, Conor McGinn.
And I got to build [00:04:00] some early mobile robots, and I got to work on some human-robot interaction, build therapeutic robots for children in hospitals. Then I got this experience where I got to go and I worked at a startup, and we deployed autonomous mobile robots to hospitals at, at a company called Kara Robotics, and there I got to actually see what happens when you put a robot into a real-world environment and, and see it do some good.
Uh, so that was really pivotal for me. Uh, I moved to the US, uh, studied at Columbia, did robotics there. And ultimately, yeah, I got to work under Matei Ciocarlie on dexterous manipulation and fell in love with general purpose robotics. So this is Ethan, and I'll just chime in here, and I guess we can even show some of the prototypes around the office as we talk.
So I had this really interesting background of being exposed to the space industry from not necessarily a space flight or a rockets point of view, but what you could do in microgravity. And I learned about, you know, growing pharmaceuticals and semiconductors and fiber optics and, you know, it's been a great twenty-five years now of R&D on the ISS, but there's these companies that wanna commercialize this.
And not many people know [00:05:00] this, but the number one cancer therapeutic on the market right now, Keytruda, benefited a lot from in-space research. You know, this is a drug that saves millions of lives every single year and generated twenty-five billion dollars in revenue just in one fiscal year, you know, from '23 to '24.
So how could I leverage space for humanity? And when I met Jamie, and he can talk about it a little bit himself, you know, his thesis has always been robotics and what robotic labor can do for the Earth and what that unlock could be. And I'll let him talk about that a little bit, but that intersection of the two is where we realized robotic labor truly only flourishes in an environment where human labor can't, especially for those first implementations.
And when you have an environment like space where it costs a hundred thirty thousand dollars an hour just to keep an astronaut alive with up-mass and down-mass and mission control, that's not even paying them or anything else involved. You know, you add launch costs, this goes up even more. That's truly one of those environments where robotic labor can flourish.
It's funny. As a founder of a [00:06:00] space company, um, I didn't know a lot about space, uh, before starting, uh, Icarus with Ethan. But, you know, I learned about a lot of this stuff, you know, this idea of, you know, things are going faster, things are getting commercialized and are going private, and I got really, really excited about it.
Before that, I was spending a lot of time working on general purpose robotic hardware, trying to find ways to make tools to collect robotic data at scale, and like many other people that were interested in general purpose robotics, looking for that one killer application, which my favorite statistic about this is kind of, you know, half of the world's GDP is labor, and if you believe space is going to be big, you have to believe that labor is going to be a massive component of it.
And that's why here at Icarus, we're building the robotic labor force for space.
What's an astronaut's time worth? Answering that question is not a perfect science, but for someone on the International Space Station, it's probably somewhere north of $100,000 an hour. That's when you take into account what it costs to keep them alive up there, [00:07:00] let alone their commute, plus a salary.
Whatever the exact number may be, it's safe to say it's on the high side of what you wanna pay for bagging up trash and unpacking groceries and other supplies. This week, we take a field trip in New York City to see some space robots that could help with this problem. I'm Jennifer Strong and this is Shift.
Now, here on the steering column is a device called Autocruise. You simply set the speed you want- Self-driving robo-taxis are already on the road in China. US cities. As the disc rotates, a mirror reflects the light in the way that depends on how the signal was recorded. This
is
the 100 terabyte hard drive. I present to you Electro, the mono man. Ladies and gentlemen. I would say that one of my greatest skills is my ability to interact with humans. This episode, we head to the Brooklyn Navy Yard to visit the home of Icarus [00:08:00] Robotics.
You gonna show me some robots? Yeah. Yeah. That's the fun part. So we like to make a joke that the company started with just two guys in a room with, you know, some cheap robot arms and a cargo bag from NASA. Hi, I'm Ethan Barajas. I'm the co-founder and CEO of Icarus Robotics. And so we were lucky enough to schmooze one of my old advisors from NASA out of a space flight-ready cargo bag.
So this bag that you see here on the table has been to the ISS. It's brought cargo up. Everything that goes to space goes to space in these bags. And these crews of, you know, four astronauts at a time will take up to seven days unpacking this visiting vehicle of three and a half tons of these same, all looking [00:09:00] similar white bags, and then repack it with down mass.
So this is all the waste, all the trash, and that takes another seven days. And we're doing this every 60 days. So to burn two entire weeks on just moving bags, you really lose out on all the science and manufacturing. I started working at NASA when I was 17 on the autonomous growth of plants for the ISS, and that was my first real internship.
And I got to see that project through all the way to its end and work on that for a few years and actually send it to the space station, which was super exciting. And then from there, I went to Caltech and I got to pursue my degree in mechanical engineering and work with some really amazing roboticists, some amazing professors at JPL, and some amazing lunar rover and Martian rover projects.
And so I got to work on this one really amazing one with Clint Carpenter and Professor Soon-Jo Chung on, uh, Chariot, and it was next-generation lunar rovers for what's called PSRs. These are permanent shaded areas or regions, and these are just craters on the Moon to find water ice. And, uh, then I was tapped for an accelerator called [00:10:00] Entrepreneur First, and that's when I was introduced to Jamie Palmer, my co-founder and CTO of Icarus.
Hey, so, uh, it's Jamie Palmer here, co-founder and CTO of Icarus Robotics. So yeah, I guess s- some of my story was, did mechanical engineering at undergrad and was really interested in motor sports. So I pursued that really far and ended up doing Formula Student, building small-scale race cars, bringing them to tracks, uh, across the UK and Ireland.
And then eventually I ended up, uh, landing a job in F1, so I spent a year at the Mercedes F1 team, uh, doing test engineering. But during my time as well, uh, before that, I got really into robotics on the side, so doing research, uh, just in my spare time for no credits with, uh, one of my professors, Conor McGinn.
And I got to build some early mobile robots, and I got to work on some human-robot interaction, build therapeutic robots for children in hospitals. Then I got this experience where I got to go and I worked at a startup, and we deployed autonomous mobile robots to hospitals at, at a company called Kara Robotics, and there I got to actually see what happens when you put a robot into a real-world environment and, and see it do some good.
Uh, so [00:11:00] that was really pivotal for me. Uh, I moved to the US, uh, studied at Columbia, did robotics there. And ultimately, yeah, I got to work under Matei Ciocarlie on dexterous manipulation and fell in love with general purpose robotics. So this is Ethan, and I'll just chime in here, and I guess we can even show some of the prototypes around the office as we talk.
So I had this really interesting background of being exposed to the space industry from not necessarily a space flight or a rockets point of view, but what you could do in microgravity. And I learned about, you know, growing pharmaceuticals and semiconductors and fiber optics and, you know, it's been a great twenty-five years now of R&D on the ISS, but there's these companies that wanna commercialize this.
And not many people know this, but the number one cancer therapeutic on the market right now, Keytruda, benefited a lot from in-space research. You know, this is a drug that saves millions of lives every single year and generated twenty-five billion dollars in revenue just in one fiscal year, you know, from '23 to '24.
So how could I leverage space for humanity? And when I met [00:12:00] Jamie, and he can talk about it a little bit himself, you know, his thesis has always been robotics and what robotic labor can do for the Earth and what that unlock could be. And I'll let him talk about that a little bit, but that intersection of the two is where we realized robotic labor truly only flourishes in an environment where human labor can't, especially for those first implementations.
And when you have an environment like space where it costs a hundred thirty thousand dollars an hour just to keep an astronaut alive with up-mass and down-mass and mission control, that's not even paying them or anything else involved. You know, you add launch costs, this goes up even more. That's truly one of those environments where robotic labor can flourish.
It's funny. As a founder of a space company, um, I didn't know a lot about space, uh, before starting, uh, Icarus with Ethan. But, you know, I learned about a lot of this stuff, you know, this idea of, you know, things are going faster, things are getting commercialized and are going private, and I got really, really excited about it.
Before that, I was spending a lot of time working on general purpose robotic hardware, trying to find ways to make tools to collect robotic [00:13:00] data at scale, and like many other people that were interested in general purpose robotics, looking for that one killer application, which my favorite statistic about this is kind of, you know, half of the world's GDP is labor, and if you believe space is going to be big, you have to believe that labor is going to be a massive component of it.
And that's why here at Icarus, we're building the robotic labor force for space.
Our very first demo was locally, could we have two leader arms and follower arms, teleoperate them, and unzip that cargo bag and take out some of... You know, we were funny with it. We got some astronaut food and freeze-dried stuff, and we put them in the correct bins that we needed to. And then as we went further, we were here in our office in New York City, and then we set up the follower arms all the way across the country in San Francisco.
This is about two thousand five hundred miles, and what we actually did is we were able to have a robot operator in New York operate those robotic arms in San Francisco in real time and unpack that same exact bag. [00:14:00] And the reason why that distance is so important is because people don't realize how close low Earth orbit is.
The ISS is only about two hundred and fifty miles away from the surface of the Earth at its closest point, and we're doing manipulation order of magnitude farther away than that closest point on Earth at low latency. And Jamie can walk through some of the magic of what we actually did to get this working and some of the next prototypes, like the V0 that you see on the air bearing table, uh, and the V1 over on the table to your left.
So here what we've done is, this was still when we were actually a team of three, so just myself, uh, Ethan, and Demetrius. And we decided, "Okay, so not many people have built a flying robot in space with two arms. It's probably pretty hard. So let's build a very early prototype and, and just figure out how is this thing gonna go together?"
So essentially what we did here was we took some, a lot of off-the-shelf components and actually built this full robotic free-flyer that you see in front of you. And here as well, we have two robotic arms. And you can see at the station over here on, on my right-hand side, you [00:15:00] can actually control this in real time to do grasping tasks.
Now, what you probably see here as well is quite a large glass table, and it probably looks a bit fancy but, or a bit funny. Uh, but what it's for is actually we have the chassis here filled with high-pressure nitrogen, and we can actually slide the surface around this table and actually fly the robot on top of it.
And what that allows us to do is essentially mimic like a 2D microgravity. So we built this rig, flew the robot on top of it, did some, uh, mobile manipulation, tested the dynamics, and then it really informed our next prototype, but also how we needed to grow the team as well to flesh it out more.
Put about 125 bar, uh, which is thousands of PSI into this small little tank, and it will give us around 40 minutes to an hour, depending on the mass on this chassis. And so this is something that we worked pretty closely with the folks at NASA on, their Astrobee team, and they helped us set this up and verify that we had everything working and our friction coefficients were [00:16:00] correct.
And so now as Jamie opens this valve and Demetrius plugs in the batteries for the manipulators and for the flight system, we'll start to see what that kind of frictionless zero-G environment looks like in two dimensions on that XY plane. There we go. Oh, very cool. As you can see, there's not really any friction here, and what that does, again, like Ethan was saying, is just mimic that 2D microgravity.
So when we fly this robot it... Okay, give it like 25% duty and just for like five seconds.
Right? And so something really interesting that you can see Is the way that the movement of the arms actually impacts the movement of the body. So when you move your center of mass forward, you actually react backwards. If you go out sideways to the left or right, you get that inverse reaction. And so one of the biggest things that we have to do is figure out the dynamics that go into this vehicle, how it moves, and the control software that goes into writing that full body controller [00:17:00] for both the arm manipulation and the six degree of freedom base itself.
You know, there's over 22 degrees of freedom in the vehicle, um, and that comes from the arms itself and just that mobile base. Um, it's modeled after something similar to an ROV. So the way that you can think about ROVs moving in underwater space, this moves through a different fluid, which is in this case air, or in the ISS, a perfect, you know, 21% oxygen-nitrogen mix.
This is the closest that we can get to it on Earth. And if you want, feel free to actually grab the controllers and see what it's like to control the robot. Yeah. Yeah. So what, what you have here is you essentially have two leader arms. So what you can do is essentially pick these up and you can move them around and you can see the small trigger here, uh, will allow you to open and close the jaw.
So if you wanna just put one hand in. Yeah. And then you can actually put a finger through here. There you go. That's you moving. Oh, amazing. Whole idea is that now you're almost embodied in a way over here, right? And you can see you have a perception of what is in front of you. You have an ability to manipulate your environment.[00:18:00] 
Now you can actually do something somewhere where you're not, and that's what we're hoping to do with space as well, is that we can now move, you know, skilled, intelligent labor in a place where it's really, really hard to get to. Embody those people, uh, very, very far away. So, yeah, you know, here to the other side of the room isn't too, too far a distance, but, uh, you know, this becomes really meaningful when you do it to the ISSA, for example, so...
We get to talk to a lot of people upstream and downstream of the problem that we're solving. We get to talk to those people that do the experiments as well. And the thing is about astronauts, there's this funny joke, you have to make experiments astronaut proof. And it's because these astronauts are super intelligent, extremely well trained, some of the highest echelon operators in the world.
But they're not necessarily a PI that's worked on a biological experiment for four years, qualified it, and then sent it to space. And so when you try to make split sec- second reactions and decisions based on what you're seeing, they don't have that. They follow their sheet, they have mission ops to talk to them, and there's this game of telephone for the PI to go talk to the mission ops [00:19:00] to go talk to the astronaut and then back.
And this idea, like Jamie's saying, of having that embodiment in space, you've effectively democratized everybody from now being able to access space where they can do their experiments themselves. And that's super valuable not only to those experimenters, but these commercial stations that can offer that as part of their services.
And that's something that was extremely unexpected for us. And I think the other thing that was really unexpected for us was when we were talking to the people that train astronauts. At the end of the day, y- you know, highly skilled operators are needed for these sorts of robots the same way, you know, construction, I guess diggers and And backhoes, you need someone that's highly trained.
And, uh, they were talking about things that would off-put astronauts that these robots are working aside. And if you look at some of the floating cameras on the ISS, they're all really cute and adorable. And JAXA, the Japanese space agency, actually did an experiment where they had a talking robot, and morale and productivity went up.
And it's this little [00:20:00] talking figurine that told people what to do. And so that's something that now we have to consider within this robot's design, and it's something that we never thought about, is the human factor. We might lose, like, a good 20 minutes if I start talking about this. But it is fantastic.
You know, even when we set up the, the first demonstration that we showed you there, people did, like, one of two things. They either lifted it up and said... you know, made a comment, like, saying it was so cute, or they went, "Rawr!" And held their hands in the air and, and make some sort of funny noise. And it's quite funny, 'cause even when people look at this robot, say, for example, they'll make comments on it like, "Oh, it looks, like, uh, really friendly."
And I say, "Well, what if it was 100 times bigger?" Uh, and then their opinion changes pretty quickly. So, you know, how these, uh, systems are perceived is really important, because ultimately, humans aren't used to having these other sort of autonomous physical agents in their space, so we have quite a gut, visceral reaction towards them.
So it's one of my favorite parts about robotics, for sure.
I think we can show you the, the comings of the next prototype, the V1. So what you saw right now was the [00:21:00] V0 prototype. We're testing everything out in its most rudimentary form, and we can show you now the upgraded version where we have stereo vision. We've started thinking about a lot more things. We can talk about, you know, the torques and how we spec'd our new actuators, and the new robotic arms that we've made, and I think Andres has them on the table there.
So Jamie, you can kick us off. Yeah. So what we just passed on the right-hand side was our new operator control interface. So you can see we've upgraded a little bit. We have some carbon fiber structures in there and some, uh, nice new motors, so it's gonna be a very high-quality dexterous manipulation coming through that.
Again, like Ethan was saying, we have, you know, upgraded perception capabilities. But I think the most interesting thing about this robot is it's really, really designed with IVA, sort of internal vehicular activities, uh, in mind. So NASA releases a lot of documentation about human factors engineering, and when you're putting a payload on the International Space Station, you have to consider all the different types of human that might need to interact with that.
So they have all these specifications of, you know, what grips, you know, a human might need to [00:22:00] have, what length their arm might be, and, and, you know, they look at the spectrum and all the percentiles of all the different shapes and sizes of humans. Similarly, we've looked at that data, and we've actually scaled this, uh, to fit all the NASA human requirements.
So the arms will mount here onto the shoulders. You have the sort of, the eyes, we'll say, uh, looking down onto the workspace, and then everything that you see has been built essentially with human factors in mind. Because ultimately, right now, what we don't want is for a station to have to make a massive infrastructure change to adapt our robot.
We want our robots to be as simple to adapt as possible, so we're building a robot to go where humans can go. Interesting. So how far away are you from putting that on the table over there? Mm. So the thing that we're actually waiting on is- The motors ... the motors. So there were a couple issues, number one, with supply chain, and the actuators are just so hot right now.
I mean, robotics is hot, and everyone are buying these actuators. And then the other thing is, like, we've gotten hit with tariffs a lot on our controllers, on the motors, and so that's always something to navigate. I think [00:23:00] we had an issue where the controllers that we just purchased were stuck in customs for a while, and so that's always a constant battle of, you know, how do you get your supply chain, especially in an early prototype phase when you're not doing things at volume, to be as fast as possible so you're not waiting on simple delays like that.
But in the meantime, we have, you know, more than enough to keep ourselves busy with , you know, like Jamie was saying, working on the perception stack. Uh, we have an entire simulation of the ISS in something called Isaac Sim, where we have our robot inside, and we can actually drive it around and interact with that environment, and that's something that we built on the back of the Astrobee team that made one for Astrobee, which is a free-floating camera that they've deployed it in 2019, and it's been up there collecting data on just movement and visuals for a long time, and many people have used that platform.
And the unfortunate thing is with the NASA reorg, um, that project's actually lost its, its funding on the NASA side, and so it's being shopped around to a couple different universities to take over its capabilities. But one thing that we're super excited [00:24:00] by is that this platform is, you know, quite modular, and it can support a lot of science And so while some folks might not care necessarily about the manipulation aspects, they might care about a free-flying robot that can help them deploy you know, new positioning algorithms and, and new rendezvous algorithms for an object like a satellite or space debris.
And so that's some of the ways that Astrobee has been used in the past and some of the ways that some people have had interest in using our platform in the future. Yeah. So I think the really exciting thing as well about having these robots in the ISS for us isn't just to populate a lot of the future space stations, uh, with these robots, but it's the idea of actually making this general purpose robotic labor that we touched on kind of at the very start.
So ultimately, we have this sort of chicken and egg problem that happens with robot learning. And, you know, we see this with a lot of terrestrial companies as well, where you need to train some sort of general purpose autonomy, which is fantastic when you have all the data, but you have to collect that data first.
For us, [00:25:00] that's particularly challenging because we have to collect that data in space. Ultimately, the goal is to get up there, and once we're up there, we're actually gonna collect, you know, hundreds of hours of data, uh, and actually begin to train these, uh, autonomous policies. And the interesting thing about these policies is there's a clear divide between new space companies and old space companies.
And people like to draw the line at, you know, SpaceX and moving launch from commercial to private. But old space and the old guard relies on a lot of traditional classical robotics, Cartesian control, predictive control, and they don't really move into this embodied AI space or intelligent robotics space.
And one of the key things for our rollout is having that human in the loop at first. You have a human behind all the decision-making. It's extremely safe. And safety is one of the biggest things when it comes to human space flight because you have these people in a metal tube in the vacuum of space orbiting the Earth at Mach twenty-five.
So safety to NASA is paramount. [00:26:00] And these commercial space stations and space companies can take a little bit more risk and learn new things. And the ISS over time has lightened up on these things, and you can actually send COTS hardware to space inside the pressurized volume. And when we have that human in the loop, we then collect that data, like Jamie's saying, and we can move to high-level autonomy, where humans are still involved, but they're choosing the high-level primitives, like pick up the cargo bag, open the cargo bag, bring cargo bag from node one to node two or to the JEM module or wherever it might be.
And there's this linearity of what we follow after we've de-risked what's safe and what isn't. And to be able to apply that intelligence that we've made with our robotics to other robots in the future, whether that be EVA, IVA, you know, cis-lunar or beyond, is super exciting to us because like Jamie said, it's moving away from just populating these space stations that are going to be few and far between for the immediate future to now being this foundational layer to all robotics in, in the space ecosystem.[00:27:00] 
So as you see people start to make moves into infrastructure, building data centers, well, who's doing the maintenance? You know, I think there's now seven hundred and twelve astronauts ever that have been to space. It's not gonna be astronauts doing what we did with the Hubble telescope and sending a shuttle and someone to do a spacewalk to fix it.
It's going to be robots. And so enabling that new future is so exciting to us. This episode was produced by me and Emma Cillekens and mixed by Garret Lang, with original music from him and Jacob Gorski. Thanks for listening. I'm Jennifer Strong