This content was created by Tim Draper. This content was uploaded by Tim Draper. This is a YouTube talk titled "The Future of Space and Satellites | Christianna Taylor" (uploaded 2017-08-03) in which Christianna Taylor addresses an audience about orbital debris, Kessler Syndrome, and using artificial intelligence and robotics for in-space salvage, servicing, and manufacturing of satellites. She presents data (citing ~1,419 working satellites, over 4,000 nonworking, and a projected 16,000 new satellites), references historical debris events (Sputnik 1957, the 2007 Chinese anti-satellite test, the 2009 Iridium–Kosmos collision), and outlines technical approaches including hyperspectral data, AI for identification and decision-making (reinforcement learning, deep learning), human-in-the-loop (level-3) automation, robotic manipulation, 3D printing and in-space smelting, and a vision for a secondary market for reclaimed satellite parts. The talk is technical, entrepreneurial, and advocacy-focused on solving orbital debris and enabling in-space manufacturing.
Chapters
So, who knows who's actually going to partner with these people when it comes to space? There's so many interesting things out there. We've had an awesome, awesome 60 years. It's amazing. But it's crowded, too. It's getting really crowded. And this is where that intersection comes into play. AI, as I said, is decision-making. AI is autonomous driving. AI is reinforcement learning, saying, how do I design something new? AI is identification. How do I make sure that that asteroid, that micrometeorite, that comet, that phenomenon is captured and sent back so we know, go look at that. Go look over there. The universe is big. And it's really easy to miss things. So AI is where we can actually go to make those decisions faster so that a human isn't up at— well, OK. Let's be real. We're all going to be up at 2:00 in the morning looking at science and space data. At least I am. I'm sure you guys are too. You just don't want to admit it. It's OK. But literally, we don't have to have a person only looking at that. I want to say Google Earth actually was taking pictures of the entire Earth, and they found a woman that had been marooned on an island in the Pacific. And the way they found it was that somebody was literally looking at that data automatically. There's entire companies dedicated to using artificial intelligence to mine through data to figure out, hey, you know what? There's some pretty cool ruins here. That's happened. We've found people that were marooned. We have found ruins. When we get hyperspectral data, when we get— Worldwide, worldwide recurring hyperspectral data. It's going to change the game because you have all the datasets. And if you remember, the hardest part about AI is datasets. Once you have that data, then you can start really making some really great decisions. But me, I'm interested in debris because, like I said, it is getting crowded up there. And I'm interested in it because the problem is that we have something called the Kessler Syndrome. Now, Donald Kessler, back in the 1970s, realized, hey, we probably shouldn't keep destroying satellites because literally they will go so fast and hit other satellites. And it gets to the point where we propagate and we will lose all— we can lose entire orbits. This is a problem because if you remember, think about the logistics, think about the data, think about the weather. All of these are huge problems if we lose access to space. And this is a problem I've been looking at probably for the last 8 or 9 years because I was like, that is ridiculous. What do you mean we have dead satellites? And when I try to explain, yeah, there's dead satellites, it's like, no, no, there is not. Space is big. We're good. No, no. Space is big, but we have a lot of dead satellites up there. So thinking about what we can do with that is where I come in. Because for every working satellite, there's 3 nonworking satellites. So imagine driving down a highway and having to navigate around 3 parked cars. And it's going so fast that if you hit one of them, all of a sudden you're a parked car and you're literally crowding up the space. Now, that's oversimplification, but that's what I am worried about. That's my daily problem. How can I do something about orbital debris? Because we're looking at a 200% increase in satellites over the next 5 to 10 years. Now, numbers, because we all love numbers. 1,419 working satellites. Over 4,000 nonworking satellites. 16,000 new ones. And I'm just like, have you heard of debris, people? Have you heard of it? Because I don't want this problem to continue. This is what it looks like. 1957, Sputnik, our wonderful beeper. 2007, largest debris-causing event in history, the Chinese anti-satellite demonstration. 2009, second largest debris collision in history— excuse me, debris event in history. The Roscosmos-Iridium collision. Those fragments are going fast. And they can hit another satellite and completely decommission it. And this is what I'm worried about. And to give you another perspective, this is us going through the altitude as to how many fragments and the density up there. You don't want to go through that type of density. Think of being sandblasted, but each particle of sand can actually completely stop your vehicle. That's what I'm worried about. But in my head, in my mindset, is that debris is actually a treasure. Now, I know, one person— one man's garbage is another person's treasure. That's what I think of, because once debris has already left Earth's gravity well, remember how launch is so extremely hard to do? It's already up there. I want to look at in-service satellite— in-service satellite manufacturing as well as servicing products so that we can actually utilize and manufacture new things in space. Right now, we have end-of-life requirements because you may be saying, all right, fine. There's dead satellites. Just deorbit it. Get it out the way. And we— and to mankind's credit, we got together in 2008 after the 2007 Chinese anti-satellite demonstration. And saying, you know what, we have to remove— either move it from the current orbit or actually move it up to what we call graveyard orbits. And I'm interested in what's in those graveyards because there's working parts. Think of it as a salvage operation. And that's what I think is so cool. I want to recycle. I want to reuse. I want to figure out new ways we can utilize those working parts, those rad-hard components. That's what I want to do. And the reason I'm talking to you guys about artificial intelligence is because I want to do it with AI. I'm literally building an AI platform that can interface with robotic arms and robotic hardware so we can figure out what can we build. Remember how I said gravity is a big issue? Imagine what can you build in space right now if you didn't have gravity? We have 3D printing from being in space right now. They can 3D print custom-made parts. We're having smelting factories from the asteroid mining companies. They can literally Melt down all that space-grade aluminum to figure out something new. That's what I'm interested in. And that's why I'm talking to you today, because artificial intelligence is powerful. That reinforcement learning is helping me design. That machine learning is helping me make the best decisions on how to navigate. That deep learning is making sure that I can actually identify what satellite is out there. And then, yes, I'm— I don't want to be fully autonomous. I want to be what you call automated, which is like a level 3 on a 5— Level of autonomy. 5-level scale. So, there is a human in the loop. Because remember, you're not taking away my cars. OK? I want to be in the loop. I want to help design. I want to be a part of this. Because space is cool. That's what I want to do with my life. My platform is literally looking at creating a secondary market of satellite parts and looking at how we can do subscription services to figure out how we can build new things. Now, I'm going to be honest with you. There's advantages and disadvantages to everything I want to do. Advantages. It's escaped Earth's gravity well. It's already up there. It's radiation-hardened. It's raw materials. It's LEGOs waiting to be played with. If you ever played with LEGOs as a kid, you're probably my friend. Disadvantages. They're old. We're talking old. Anyone ever try to get a file from an Apple to a PC in the '90s? How do you make the compatibility? That's an issue. And I don't know the answer to that. And that's why it's so exciting because there's AI to help me find that answer. Compatible. Low reliability. How many people get frustrated with an old computer? You buy a new one, you're like, how did I ever deal with this? How? GPUs are changing the game on computers right now. So how do I make sure the reliability of anything I want to build works? How do you make sure that you don't have to rebuild everything, every component from the ground up? And I don't know the answer. And my hope is that artificial intelligence is going to help me find that answer. There's 4,000 satellites of raw material up there. 4,000. We think about this option, deorbit it or salvage it. I'm interested in salvage. Sometimes, yes, you absolutely should deorbit things. But I'm personally interested in the salvage operation. And that's what I'm looking forward— that's what I'm thinking forward to when it comes to artificial intelligence. Raw materials. Some creativity, some imagination, and engineering. A human can do all that. Artificial intelligence can make it more efficient. It can do it faster. And it can think in terms that we haven't had a chance to think in because we have only our own biases and our own experiences. And that's where I think the innovation is. So, yeah. I want to make the Enterprise. But I'll probably wind up making Firefly, if you know the reference. Because Firefly was made out of junk parts. They literally flew by the seat of their pants and fixed everything along the way. But you know what? We have 4,000 satellites. We got time. And that's what I'm interested in doing. This is my company. Feel free to tweet me, talk to me, and any questions? Yay! Yay! Yay!
