[00:00:00] Host: Welcome to Lockheed Martin Space Makers, the podcast that takes you out of this world for an inside look at some of our most challenging and innovative missions. My name is Ben, and I'll be your host.

[00:00:14] In season two, we explore Lockheed Martin's bold new vision of a future we call "Space 2050." We partnered with our Advanced Technology Center to bring you an inside look at the innovations and technologies we are developing to make that future a reality. Because getting there is just the beginning.

[00:00:35] From distant outposts on Mars to colonies living on the Moon and smart cities here on Earth, we will need artificial intelligence to help make it all work. AI will shape our future and may even fundamentally change how we work and engage with society. My colleague Natalya Oleksik takes a closer look at why AI will play a critical role in everything we do in space, from human colonies on the Moon to trade routes to Mars. My colleague Natalya Oleksik takes a closer look.

[00:01:09] Natalya: So I'm sitting down with Dr. Steve Gerali. Steve, could you tell us your name, your name and your title, please?

[00:01:16] Dr. Steven Gerali: Sure. My name is Dr. Steven Gerali. I, my title is a chief architect for our space BITEE group, which stands for Business Innovation, Transformation and Enterprise Excellence. It's part of our kind of IT innovation arm.

[00:01:29] Natalya: We're here to talk about today's transformation and where you're taking that for the future of space. And everything that you are working on where we anticipate revolutionary changes to how we not just go there, but work up there and develop a new space economy. Can you tell us a little bit about the hurdle and integrating artificial intelligence into our space exploration and once we overcome that hurdle what it means for humankind.

[00:01:54] Dr. Steven Gerali: Yeah. I, you know, I think the biggest hurdle for any new technology is how well do we adapt it. And how do we integrate it with our human counterparts, where it actually acts as a way of helping the humans do what they need to do and do it better. Um, and not necessarily always taking all the actions on behalf of the human. So, uh, oftentimes people think of automation and they think, "Oh, we gotta automate the job. It's gotta be 100% computer."

[00:02:16] And, "Oh, we gotta find ways where we're symbiotic. where both the human and the computer can work together to get the best out of what both offer in the business as a whole." And so, uh, when it comes to the whole AI side it's really, what are the right level of use cases where we do a lot of manual effort where we can help automate that for the end user, where they don't have to do the things that they don't want to do.

[00:02:39] So oftentimes I use the term practice versus process. If you think about a doctor's office, right? A doctor's there, they have their practice, right? They're there to have patients. They're there to help the patients. But there's a process that they have to go through in order to make that happen. They have to be able to work with insurance companies. They have to be able to work with providers. They have to work with pharmacies and so on and so forth.

[00:03:00] That's not necessarily all the work that they wanna do. They really wanna focus in on the patient care and what they gotta do to help the patient. It's the same thing with the folks working in space. They wanna focus on the science. They wanna focus on the important things of the mission and not have to do the mundane things that just don't take a lot of... Like, they take a lot of energy, but they don't actually produce a lot of value. And so it's finding that right level of engagement that we wanna have with those teams.

[00:03:24] Natalya: What kind of technologies are we developing here at Lockheed Martin that based on AI that will help do that?

[00:03:30] Dr. Steven Gerali: So some examples here are, uh, some of the telemetry processing that we're doing. So oftentimes if you think of telemetry coming off the spacecraft a lot of data, very voluminous, a lot of velocity. And oftentimes it's, well, how do you find the needles in the haystack? How do you find the things that, "Ooh, this is important. I should pay attention to it." Or, "We have a problem here. We should go look into it more." And so what we're doing is we're developing AI solutions.

[00:03:54] There's an effort that, an umbrella effort that we call [Tetori 00:02:47], which basically says, we're gonna create a set of algorithms that are gonna help our operations team in mission, basically pinpoint anomalies, pinpoint issues, help with tracking, help with isolation. Help with effectively finding that needle in the haystack and being able to tell you what's the root cause of it. Is it 'cause we don't have enough power? Is it 'cause we don't have enough propellant? Is it 'cause we don't have enough whatever the answer is, right?

[00:04:20] And so, uh, a lot of these algorithms they... There's different types of machine learning algorithms. The one that we use is an unsupervised machine learning algorithm. What we do is we feed it nominal data, how the system should be behaving under, under normal conditions. And then the algorithm itself starts to pick up on anything that's abnormal. And over time can effectively tell the end user, "Okay, this is a anomalous. This isn't what we typically see."

[00:04:45] And, "Oh, by the way, we saw this thing go off and these other 10 telemetry values go off as well." So somewhere within that realm, we got something that's not right. So we can at least pinpoint out of hundreds or thousands of telemetry points what are the things you should be looking at. And so we're building out algorithms to kind of support that level of effort.

[00:05:04] Natalya: Those algorithms will be essential for Artemis missions and Orion, essential for protecting astronauts lives and essential for eventually helping us build colonies on the moon and beyond, how trustworthy are they?

[00:05:18] Dr. Steven Gerali: So there's a lot of things when it comes to trust and AI. if you feed it bad data, it can develop a personality of its own. So for instance... Oftentimes, I think this was actually an Amazon's case. They were helping the police departments determine criminals or bad folks. And they oftentimes would give pictures of only a certain character type of person. Maybe a certain race, a certain sex, whatever what forth.

[00:05:43] And what ends up happening is the algorithms don't know any better. The way that we program these AI algorithms is different than we program a system. Typically in a system we're setting the business logic, the business rules and all the underlying en- endpoints. With AI, we're actually setting the algorithms with training data. So if I give it a bunch of data and it's all men and it's all White men, the algorithm's gonna be more poised to identify White men than anything else, right?

[00:06:09] So the question really becomes, how do we make sure that our data is broad, it's diversified so that it actually includes all the different possible conditions that we kind of run into, so that we can build the trustworthiness. And I'll give you a good example. In our supply chain group if you think about Lockheed Martin Space, 70% of the cost of the makeup of the satellite is built in our supply chain.

[00:06:31] Now here's the issue, we have to take a lot of parts in. And in order to do all that system integration, we gotta know when all the parts will arrive. And so we were building out machine learning algorithms that effectively would predict the lead time, how long it would take for the parts to come in. And we had to build trust with the folks that were ordering the parts that they'd feel comfortable with what we were giving them.

[00:06:53] So if Amazon tells you it's two days... Sometimes it's two days, sometimes it's a week, sometimes it's kinda all over the place. It's the same thing with the parts that we get. So the question becomes is, how do we gain that trust for the end user? And so they wanted to be more involved in that process. So what we end up doing is kind of a man in the middle where we provide predictions and then we ask the end user, how accurate is that prediction?

[00:07:16] Did we hit it? Did we not hit it? You know, what do you think, uh, things look like? In fact, you'll see that in a lot of AI systems out there today. Like, if you use your Alexa and you ask her a question, she'll say, "Did I get it right? Was this helpful?" And it's really just trying to help build that model for, "yes, this is good data. This was a good result." So that we can keep tagging and promoting the right data and build what people would consider to be a truthful model of what real really is.

[00:07:43] Natalya: Is the system of AI ever perfecting? Is there ever, are there ever enough data points to make you feel comfortable that our programs are protected, our astronauts and space are perfected? Or is an ever perfecting process?

[00:07:58] Dr. Steven Gerali: Yeah. It's definitely a process. It's something where we get better each and every day. But we're not on the order of being 100% perfect in anything that we do. We are human. We do make issues. We tag things wrong. We train things incorrectly. We do a whole host of things that could always be better. The question is how do we get more and more along that line of accuracy where it's, it's good enough to meet the mission's need.

[00:08:19] And it's confident enough that even if we were to get it slightly wrong, it doesn't cause harm or issue to the astronaut or others around them and kind of moving forward. So definitely kind of a growing space. And oftentimes we talk about the accuracy of our models. So we will predict what our models should be. So in that lead time example I gave you, we'll say it's gonna take you 10 days to get in. And if it comes in 11 days, we're a little bit off. "Well, okay, let's figure out how off we were."

[00:08:45] And then over time, you know, the goal is how do we get and better at those predictions? How do we... For certain providers, we know they're gonna deliver relatively on time. They're gonna be stable. It's gonna be good. For others, it may be wildly different. And it may be different based off the parts. Some parts are more complex than others. So sometimes a very complex part may take a lot longer for it to come in versus a more simpler part.

[00:09:08] So anyways, our goal is really just to work on the algorithms, get them to provide as best accuracy as we can. We don't really... We're not shooting for 100%, but we're shooting for as accurate as we can get in order to deliver the result for the mission.

[00:09:21] Natalya: So not shooting for 100% on the ground supply chain, never able to guarantee 100% in space human space flight, human space exploration, what do you do with that uncertainty? Is, do you ever anticipate getting 98% certainty or, or is that all part of ex- exploration, which is taking a chance?

[00:09:41] Dr. Steven Gerali: Yeah. I, I mean, at the end of the day when it comes to humans, you want it to be as sought as you possibly can, right? And certainly if there's any failure that you have redundancy in those failure conditions that you can support the mission, support the astronaut, support whoever you're helping there is always gonna be a certain amount of data.

[00:09:59] You asked about, well, when you have lots of data and it's voluminous and it's coming at you, how do you deal with it? And in some cases, if the data is so large and you're not processing it fast enough, it's already too late. It's already passed. There's nothing we're gonna do about that. Gotta move on to the next one and the next one and the next one.

[00:10:15] So really what it comes down to is, what's the minimum amount of data that we need in order to make the best prediction at the best time and give the best result that we can to whoever needs it, whether it's an astronaut, whether it's an operator on the ground, whether it's anybody, you know, trying to support the mission.

[00:10:32] Natalya: How do we make strategic decisions with data? You talked about telemetry, what other types of data can we get from space that help us make strategic decisions?

[00:10:42] Dr. Steven Gerali: Yeah. So I'll give you another good example. You know, in the space of IoT, of internet of things we have sensors. These sensors are picking up things. And typically you have actuators. These are actually doing something based off of some sensor that they're picking up on. So for instance, I always tell people think about your house. You have your thermostat, it's reading the temperature. When the temperature gets below a certain level, okay, we want to turn up the heater.

[00:11:06] Or if it gets too high, we wanna turn off, we wanna turn on the, uh, cooling. And it's the same thing here in the space side. We have sensors that are actually picking up all sorts of different things. It could be imagery. Like, for instance, we have imagery of, of solar flares that are happening with the sun, uh, and they happen all the time. But there's certain ones that we're looking for and that wanna see.

[00:11:26] So rather than a human having to sit there and watch hours upon hours upon hours upon hours of video, if we can train an algorithm that can actually do that on their behalf, find the ones, the segments that we really care about and then be able to pinpoint where in those videos they occurred. Now we're saving that individual lots of time, where they can go, "Hey, I know there's a solar flare here at minute five, minute six, minute eight." And they're skipping all the rest and all the stuff that doesn't matter to them.

[00:11:52] And really it's, it's about when we have all these sensors at our disposal, we get overwhelmed with all the data. That happens, you know, in space. It also happens when we get into battle management systems. And we're trying to, uh, isolate and figure out, okay, what's really happening in this given space. And so we have examples like that. We also have examples like, where we are looking on earth with, like, pictures of what's happening on the planet.

[00:12:20] And in fact, we've done applications for a group called Help Now, which effectively was, how do we take a look at disaster recovery initiatives with the red cross where maybe a bridge is blown out and we can't deliver food or water or whatever, humanitarian relief to the folks that need it. The more that we can kind of look at all the sensor data that we have, whether it's digital data, whether it's imagery, whether it's any of these things, make sense of it and redirect our assets in a way that's actually gonna help the mission, that's really what's gonna help win the day.

[00:12:53] So in the example of Help Now, we actually, you know, worked with the Red Cross. We developed an application that effectively took a lot of the imagery that comes off of our satellites, helps them to plan the future for how they're gonna attack disasters and then use that information to then go and execute. And it also includes, well, how do you plan and manage your resources, right?

[00:13:17] So we have human resources. We have cars and trucks and all these other things. There's only so many more resources that we have, where do we apply them to get the best optimization out of them? So what are the high value targets? How do we optimize around that and kind of support it? So I would say there's, there's stuff in space that we deal with. There's stuff on the ground that we deal with. There's all these different mission components. But at the end of the day, it's how do you make the mess sense of it optimized to help whoever the customer is in solving their needs.

[00:13:47] Natalya: Will that always require a ground based station? Or is there a way that astronauts could be trained to be the key decision makers based on the data?

[00:13:57] Dr. Steven Gerali: So I think what you're seeing here, especially as we go to Mars and other places, right, is that the amount of time that it takes in order to send data back to earth, to get processed, to deliver results it, it's not gonna, it's not gonna work for us as we get further out, right? And so that's where edge processing really kind of comes in. So some of the things that we've been working on is, how do we cleate, create a constellation of satellites that effectively can communicate with one another in a swarm like effect? Meaning, if we need more computing power, they can join the swarm. If we don't need as much computing form, they can drop off the swarm.

[00:14:33] Natalya: Or go to sleep or rest.

[00:14:34] Dr. Steven Gerali: Or go to sleep. Exactly. And the whole point of it is for a lot of those astronauts, they're gonna have to do processing at the edge. We have to be able to provide them the compute infrastructure that they need when they eventually get to Mars. And be able to kind of orbit around and give them answers that they require. 'Cause if they have to send it all back to earth, it's gonna take too long, too much time, too much energy, too much effort. We'll never get there.

[00:14:56] So we have to find ways of how do we empower them when they're out there and give them the infrastructure that they need. We're paving new roads, right? Jeff Bezos would always say, when he was setting up Amazon, he could sit on the backs of giants, right? He could literally... He didn't have to go and invent FedEx or UPS or any of these things, right. He can use them in order to deliver what he needed. That was already figured out. He didn't need credit card systems or payment systems to manage that. It was already there.

[00:15:21] Natalya: The infrastructure was there.

[00:15:22] Dr. Steven Gerali: Right. So the question is, how do we build the infrastructure for the astronauts in Mars, on the lunar components, on all these different areas to give them the compute that they need, the sensors that they need, the stuff that they need to be successful? And if we have time dimensions that are relatively small, we have to get to more and more of that edge processing to give the power to them right then and there so they can make effective decision, and not always have to ship it all back to us.

[00:15:47] And we have to be very particular about the data that we're looking at. How do you get the least amount of data that's required in order to do the job? Because like I said, you can become overwhelmed with all the data that's out there and go through analysis, paralysis and not make decisions and everything else. So it's really kind of just focusing, you know, where you want to put your energy and your effort.

[00:16:08] Natalya: You talked about paving new roads and you talked about swarms of satellites, do you see a time at which point they will become autonomous enough to pave their own roads up in space to be, for example, "A space craft is, or space a ship is here, here are where the humans are. Let's go to them."

[00:16:25] Dr. Steven Gerali: Oh, absolutely. I mean, we're already close, right? So you have computer vision, which effectively is a way of taking images and video and being able to identify what's in that image or that video. So today we actually do that for some of our production operations. So think of our high bays where we're actually building out the satellites. We have folks... There's only, can only be a certain amount of folks that can actually be on the platform.

[00:16:48] So we actually have cameras set up to watch and make sure we have no more than a certain amount of folks, or we notify people. It's the same concept. We'd be having some type of imagery being collected by the satellite. We can adjust the actuators of that satellite to go to different sections, different orbits, different locations, different whatever, to follow the tracking and do what we needed to do in order to meet the mission.

[00:17:12] And so the end goal really is, you know, how do we prepare the algorithms in such a way that we can plan for how we wanna execute that resource at a given time? So ab- absolutely. Can we... Uh, I mean, we already have autonomous driving. I mean, look at what we have with, uh, Tesla and other cars.

[00:17:30] Natalya: Mm-hmm [affirmative].

[00:17:30] Dr. Steven Gerali: Where literally the edge processor is sitting on the compute- o- the, the edge processing is sitting on the actual car, looking at imagery in real time making decisions. And mind you, this isn't... I mean, driving is a complex system, but there's only like four, you know, rules of the road, right? It's-

[00:17:47] Natalya: Don't hit other people.

[00:17:48] Dr. Steven Gerali: Exactly. Don't hit other people. Use your, use your indicators to indicate which way you're gonna go. Make sure you stop. Make sure you follow the speed limit. Like, it's, there's not a lot of rules to it. But at the end of the day, it works, right? And we have to develop those same sets of rules when it comes to how do we want the satellites to work, to function, to manage and what mission are they going after. And how can we adapt them to meet that mission?

[00:18:11] But you're talking about two different things in some ways. And that is when you say there only a couple of fundamental rules about driving on earth that incorporates a human brain with human intuition. And what your talking about in space are swarms of satellites, different missions perhaps. One of them might be to support an Orion going to Mars and another one might be to support a go satellite bringing d- data back to the ground, but there's no human intuition involved. Is that concerning? Because human intuition often fills in the gap between what you know and what might happen.

[00:18:43] Well, so that's where I think you gotta get the best of both worlds, where you gotta interact the humans with the algorithms and, until you build the confidence. And eventually once you've built the confidence and people go, "Oh, yeah, this algorithm 98% of the time it hits the mark." I don't need to pay attention to it anymore, right? And even if it were to miss, here's what the effect of that missing is.

[00:19:06] It doesn't mean we're gonna lose the asset. It doesn't mean we're gonna lose the mission. It doesn't mean any of those things. It just means, "Okay, we're gonna have a little flubber here. It's not the end of the world. We'll learn. We'll move on." And, you know, I think over time we just work to identify which tasks should be human based, where the intuition is absolutely needed and vital and supportive.

[00:19:26] And which of those things are things where, well, you know, the computer can figure it out. Like tactically the computer can go, if we're driving. "Okay, I know what the speed limit is. I know there's nobody in front of me. I know there's nobody on the sides of me I can go forward." Okay, fine, go forward, right" And, and don't get me wrong in any of those systems today, the human is actually the intuitive part of that Tesla driving system.

[00:19:50] They don't just say, "Oh, no, no, just go ahead and let it do its own thing." We've had people do that and it makes terrible mistakes. It's harmed people. It's harmed property. It's harmed lots of things. And the humans are there to try and add the intuition but the problem is when the human turns off.

[00:20:04] Natalya: Yeah.

[00:20:04] Dr. Steven Gerali: And the human goes, "Oh, well, I got that notification three or four times and I didn't have to do anything. So now I'm not gonna pay attention to. And in fact, I'm not paying attention to any of the notifications you give me. I've just been, you've alarmed me too much. I don't care." And we're just gonna go forward. And hopefully 98% of the time will get there, but the 2% can be bad.

[00:20:23] Natalya: Mm-hmm [affirmative].

[00:20:24] Dr. Steven Gerali: And it's because we didn't think through that whole problem set of, okay, how do you not, you know, alarm the person so much that they just turn off? And now they're not even involved in the whole solution. It's kind of like checklist. If you think about checklist, what does it do? It turns off the brain. Why does it turn off the brain? All I gotta do is go through my checklist. "Oh, checklist one, two, three, four. I did them all. I'm good."

[00:20:44] Well, it's fine when you wanna perfect process. And, you know, the process is solid and everything else. But a lot of the stuff that we do in space, yes, there's process, but there's some amount of experimentation. There's some amount of exploration and other things that we're not gonna really know until we get out there and we try it, which means that the human still has to be involved in all of this. And we gotta know that, you know, you can't just always turn off the human element. The human element is actually can be quite helpful to the overall system.

[00:21:11] Natalya: So what you're saying is data's not gonna solve all, all our problems. It's important to have people up there.

[00:21:16] Dr. Steven Gerali: Yes.

[00:21:17] Natalya: And of all the technologies we're developing, all of the data, AI technologies we're developing to support that, are there any that excite you more than others?

[00:21:27] Dr. Steven Gerali: Oh, there's, there's a lot that excite me.

[00:21:30] Natalya: Let's talk about a few of those.

[00:21:32] Dr. Steven Gerali: Okay. [laugh] So I'll be honest with you when it comes to the power of data, we've not even come close to it, right? Are just at our infancy of being able to collect, store, understand, interpret and take decisions, proper decisions against all that data. If you think about it, when we fly an aircraft... Let's say the F-35, for example, uh, what do we do? We slap a hard drive in it. We fly the plane from point A to point B. When it comes back, we download all the data off the bird 'cause it's so much data. And then it takes a long time to process.

[00:22:11] Well, none of that stuff is in real time, right? And one of the things that we've been talking at about... And actually there's a great book, it's called The Kill Chain and it's by Christian Brose. And it's really, how do you go from sensor to actuator, to actual attack and actual elimination of an enemy component, right? And they refer to that time as the kill chain.

[00:22:32] From the time that you identify something to the time that you actually take some and you've actually neutralized the problem. And they wanna try and get that time to be smaller and smaller and smaller. Well, as we all know, sometimes the data's not good. And when the data's not good, we make really bad decisions. And when we make those poor decisions, it has big impacts both to civilians, to all sorts of folks. But one of the things that it's teaching is this concept of net-centric warfare, which we've had for 10 or 20 years.

[00:23:00] But really what Christian talks about in his book is often the customer wants to buy the next platform and the next platform. So I go from an F-16 to an F-22 to an F-35 and I'm just buying bigger and bigger platforms. But really the power is in the data. It's in, how do I integrate all these systems? So my F-35 talks to my satellites, which talks to my ground stations, which talks to my battle systems, which talks to my troops, which talks to all these folks. And we refer to that as joint operations since part of our 21st-

[00:23:27] Natalya: Between them-

[00:23:28] Dr. Steven Gerali: Yep. Yep. It's part of our 21st century war fighter, right? And it's, how do we take all of that data and make sense of it so that the war fighter has exactly what they need when they need it, and that they get in an amount of time that it's actually can be useful, that they can do something with that data? Oftentimes, sometimes the data comes in late. It's already too late. We can't do anything with it. It's useless.

[00:23:50] Other times there's so much data. We can't capture it all. We just go after the big things, the big hitters, right? And so I think the, the best thing that we can do is continue to work on our data dexterity and how we collect the data, manage it, synthesize it, understand it, take actions on it and continually improve upon that. And I, in the kind of space as a whole, they refer to that as the data fabric.

[00:24:13] How do you bring all that data together, provide the tools, the algorithm and... When we talk about AIML, there's this whole concept of the marketplace. Which is to say, the same way you buy an application on your phone, it's the same way you can buy an algorithm from an algorithm marketplace. So if today you want anomaly detection, great, here's an algorithm for that. If tomorrow you wanna do predictions on something, like predict my lead times or predict whatever or my defect rates or whatever, okay, here's an algorithm for that.

[00:24:43] And so what ends up becoming is this whole ecosystem where folks can very quickly build end to end solutions for their end customers. But if you have garbage data, you're gonna get garbage results. So you gotta have really, really good data. You gotta make sure that you use the right algorithms for the right job. And that we have ways of combining all of that data to get a single integrated picture of what's happening.

[00:25:06] So we can make the, the war fighter know, "Hey, this is your best option." And at the end of the day, the war fighter still has to use their brain. They still have to go, "That makes sense to me. I see what's going on. Yes, let's go forward."

[00:25:21] Natalya: You're talking about data as a service and that's for defense and keeping our customers secure, but also as a service to help earth, correct? Bringing it down to earth to help us with crop prediction, maybe biometrics, things like that. All of that takes processors and where are they in this mix?

[00:25:40] Dr. Steven Gerali: Yeah. Well, as far as the compute goes, the compute is kind of all over the place, right? We have compute in public cloud providers like Amazon, Azure, GCP, you name it. There's different kind of cloud vendors out there that provide the level of compute that you need, that you can grow on demand as you require. There's stuff that we have on premise that we deploy ourselves, that we manage ourselves.

[00:26:02] And it's the combination of that infrastructure that we use to kind of pull in the data that, that we're looking at. So, for instance, if you're looking at... We talk about it as helping mother earth, but effectively we wanna know where do we wanna plant all of our plants and other types of things. We oftentimes will use these processing units in order to kind of say, "Okay, well, how much precipitation do we get? What's the cloud cover look like? How much sunshine? What's the average humidity, the temperature? What, how do these crops typically, you know, work in a given region?"

[00:26:34] So then we can basically say, "Okay..." And oftentimes you gotta circulate your crops kind of all over the place. And, and not only that, but then you also get into the whole aspect of when you get into cities of farming, vertical farming 'cause now we don't have a lot of space just on the ground. We gotta create vertical components where we can actually grow things in buildings and along...

[00:26:54] I... It was interesting. It was actually at, uh, the D- Detroit Plant for Ford. It's called Greenfield. And anyways, if you look at all their buildings, at the very top of the building they have a bunch of plants. A bunch of like just... You know, "Oh, we're growing corn over here. We're growing other things over there."

[00:27:11] Natalya: Soy beans or...

[00:27:11] Dr. Steven Gerali: Yeah. And we're asking, "Well, why are you guys doing that?" And they said, "Well, actually it keeps the temperature of the building lower, right?" And not only that, but then we also get the benefit of, you know, we, we get the crop off of it as well, right?

[00:27:22] Natalya: Food.

[00:27:23] Dr. Steven Gerali: And so what I would tell you is I, I think we gotta think of different ways that we attack problems. And sometimes they may be a little bit different than what we expect. We oftentimes kind of think, "Oh, well, no, it, it's gonna have to be in some rural community that's where we're gonna grow all of our crops. And there's no benefit of trying to do it in the city." And it's like, "Well, no, think about other ways where large..."Like if you go to Singapore or other places, they grow all their stuff. They just do it vertically, you know, in their given regions.

[00:27:50] And so what I would tell you is that when we're trying to process the problem and trying to break it up, we'll need some level of compute to make that happen. And whether it's in our own data centers, whether it's using public data centers, whether it's using shared data centers from other teams or other groups or communities, we have to kind of bring that all together and integrate it and make sure that we can kind of show what that end result's gonna look like and be.

[00:28:18] Natalya: And all of that begs the question, security. Data is a universal language, correct? So we have this proprietary data that helps allies and customers flourish in their goals. But currently we do live on a contested world and how do we keep this data secure? What do you see in the future for that?

[00:28:37] Dr. Steven Gerali: Yeah. So there's different types of security that we have out there today. So we have security for data that's in transit. When it's going from one system to another system. We have encryption, which is for data that's at rest. And, you know, when you think about the security of these things, we have to know that the parts and materials that we're using on spacecraft are authentic. They're not fabricated parts.

[00:28:59] They're not things that are gonna hurt the astronaut or other folks. 'Cause there's a lot of fake parts or counterfeit parts that are out there. And so there's work that's happening a lot in the blockchain area in order to help with our supply chains, so that where that part has been, all the different facets that went into it, what parts were used, who manufactured them, who tested them, when did they do it.

[00:29:22] And it's basically creating this digital ledger of all the transactions that happen against that part. So by the time the part arrives to you and you install it in your finished product, you can feel confident, yes, it hasn't been mali- manipulated, changed, any of those things. We feel confident that part is really what that part is. And then on top of that as we're collecting data off of it, we're saving that data. We're protecting that data. 'Cause at the end of the day, you know, we don't...

[00:29:50] Well, I should say there's some data that we care about that's part of our intellectual property. For those things that we wanna be able to store and protect will do that. There's other data out there that we produce for NASA and others where they want to share the data. They wanna make it available to anybody and everybody, whether it's imagery, whether it's anything else.

[00:30:07] In which case, you know, we gotta provide mechanisms so that they can share that with their community members. So no matter what, there's some cases where there's a need to share it. There's some cases where there's a need to know. Sometimes there's cases where it's a need to protect. And so you kind of have to look at what each of those kind of problem areas are and kind of go, "Okay, well, which one do I need to use here? Which tool am I using to meet the need?"

[00:30:30] Natalya: And so let's fast forward to the future and you have swarms of satellite that are following space ships as they go to Mars or they're tending to structures on the moon and helping repair buildings, things like that and they get hacked and they no longer work, how do you protect against that?

[00:30:46] Dr. Steven Gerali: Yeah. Well, so I'll give you a good for instance with our factories that we have today. So we have our intelligent factory effort where we're actually bringing online digital equipment that does both the manufacturing and the quality checks for any of the parts that we build internally ourselves. And you have to have some way of being able to quality, verify and check to make sure that it actually built what you wanted, right?

[00:31:11] And so what we typically do is we isolate that environment from everything else in, in the network so that we are lowing the attack surface. So if somebody wants to try and attack it, okay, well, we've very specifically restricted it to just one spot, one area, one level of effort in our network. And we typically firewall these things and make sure that only, only the folks that need to connect to them can connect to them.

[00:31:37] Now, if, if somebody were to hack in and actually get access to that unit and they're manufacturing parts, could they add something else? Well, absolutely. And if you have no way of validating or verifying that the finished product meets the need, that they didn't like, you know, totally rearrange the design and now there's a structural defect and now we're not gonna be able to hit some of our thermal projections or some of our weight restrictions or whatnot then, yeah, we have a problem.

[00:32:05] And so oftentimes it's just as important in the way that we manufacture things, is the way that we test their quality and very verify those things. And you have to have a full cycle system that lets you know, "Hey, we created it, somebody validated it and now we can apply it." And making sure that whoever has been in the system, you know, who did it, when they did it, why they did it, all those things.

[00:32:26] And if you see things that, that look wrong, then we have to go in there and go, "Well, did somebody attack the system? Did somebody come in and adjust or change?" And if you think about the same stuff that we do here on earth for manufacturing, we're gonna have to do up in space. So whether we're on the moon and we're taking, you know, material off the moon or off of an asteroid or anything else to build parts and material, we're gonna have that same mindset.

[00:32:53] Uh, and a lot of those systems are gonna have to be closed. So where today, you know, they just sit in open air and they're in a factory and they work fine. You put them in space, they're gonna operate completely differently. 'Cause people didn't plan for it being a space based manufacturing component. So we have to figure out, "Okay, well, can we create a closed system where these things can work. And it, when they have problems, are they serviceable? Which components are servable, serviceable? Which ones aren't? How do we recycle those components? 'Cause, you know, it's not like we're-

[00:33:23] Natalya: Sustainability-

[00:33:23] Dr. Steven Gerali: Exactly. We don't have a, a huge landfill where you just pile up on the moon and say, "Hey, guys, here's all the junk that we've produced today that's no good." You know, we have to think through all of the logistic of how we're gonna run that space economy, how we're gonna move folks. And not only that, but a lot of folks are talking about, well, how do we get out to space cheaper, quicker, faster, right?

[00:33:44] And certainly, you know, with reusable rockets, that's great. That's bringing the cost of launching down. But we have to do it even more. So people talk about space elevator and being able to get things in a low earth orbit for relatively cheaply. And then you, it's almost like a gas station you fill up in your LEO spot and then you fly off to wherever you need to go, right?

[00:34:03] And so there will be this whole ecosystem of a space based services that are gonna be required in order for us to get from the moon and beyond just commercially, which is already happening. NASA's already kind of said, "Hey, we, we expect the commercial entities to kind of manage that. We're going after the bigger fish. The, the really stuff that's out there like Mars and others. And let's focus in on, you know, deep space." So that's great.

[00:34:29] So now we've got commercial entities working on lower earth orbit and selling services that you've got startups that are launching little small sats. They're selling all of their imagery for relatively cheap, cheaply 'cause that's all becoming more and more commercialized.

[00:34:43] Natalya: Back in the day, it wasn't so cheap. [laughs]

[00:34:45] Dr. Steven Gerali: Right. Exactly. [laughs] and now people are sharing time on the rocket. Like, "Hey, I need to do my payload, you do your payload and we're gonna launch both payloads at the same time or multiple payloads." And so it's interesting. But I think that the main conduit is that in the space economy like we talked about before for infrastructure, we gotta have all the infrastructure there for people to use, whether it's propellant they'll be required, whether it's to gas their vehicles, whether it's materials that they'll need to build what they need to build either on the moon or other places to really kind of keep pushing that boundary of exploration.

[00:35:24] Natalya: Do you see a time when that will happen?

[00:35:26] Dr. Steven Gerali: Uh, you know, it's interesting, we, we're working on like the 20 year and 30 year plans for, like, what do we think a space could look like. And, and there's a lot of folks in, in this area that look at, "Oh, well, you know what? We could actually ride asteroids for a while." And it's like, "Well, that's interesting." They actually think that they, that we can actually produce gravity with an asteroid that's spinning fast enough to give you that whole aspect if that's what you need or desire. I think you're seeing right now is you got folks that are working on comms for the moon. So the same way you get your Verizon or your AT&T-

[00:35:58] Natalya: Parsec.

[00:35:58] Dr. Steven Gerali: Yeah, you'll have that set up on the moon. So you can have your comms ready to go and doing whatever you need to. And so I think what we're seeing in the space as a whole is that everybody's seeing the need to, "Okay, well, how do we build out the infrastructure and what areas do we all specialize in?" And everybody's got their own little niche. You got some people that their niche is just the rockets.

[00:36:19] You got others that it's the satellites. Or others that it's, you know, different mission components or robotics or other types of items. And ultimately, I think... It was funny. I was telling my son, I said, "I hope at some point you actually get to the moon, then I'm sure they'll even have tourism." I mean, we already have that, you know, today with what you're seeing with Blue Origin and SpaceX, where folks are going out... And Virgin Galactic, where folks are going out into space as a form of, "Hey, this is interesting. It's different. It's not what we're expecting."

[00:36:51] And I think what we're gonna end up seeing eventually is that you may be able to go and visit the moon and there'll be a whole museum there where they laid the flag down and you can see everything about it. And we'll be able to run drones and cars and everything off the moon doing our own explorations of things. Not just ones where it's on behalf an asset, but ones that are commercial oriented, meaning I can rent time on there and I can go be an explorer, right?

[00:37:17] And people wanna have that kind of ability to see what's out there. And, I mean, with the amount of population growth we, we're having we really, we have no choice. Space is the next frontier. So we have to treat it with the reverence that's required, which is to say, we gotta find, you know, the next planet, the next big thing to help grow our capabilities, grow the opportunities for folks. And, and I, I, I tell people this all the time 'cause they always go, "Well, space is hard."

[00:37:46] Yeah, space is hard but here's the thing, the impact is immeasurable to humanity. If we get it right and we do it right and we can actually explore things... I mean, I think back to, you know, Star Trek back in the '70s and '80s, I can talk on my watch just like they could talk on their transmonitors, right? I can literally ask Alexa to go and, you know, run programs and do analysis for me, right?

[00:38:11] Like, the idea of us interacting with computers... In fact, I was reading a Gartner study the other day. They were saying that people actually talk more to their digital assistance than they even do their own spouses.

[00:38:22] Natalya: [laughs]

[00:38:22] Dr. Steven Gerali: And I'm sitting there going, "Wow, that's crazy." But sometimes that can be true, right?

[00:38:26] Natalya: Right.

[00:38:27] Dr. Steven Gerali: 'Cause, uh, uh, wherever I go I, I have some digital component in my pocket, in my car, in my work in my home in my whatever that it's just always readily available. It's hard not to, right?

[00:38:40] Natalya: Mm-hmm [affirmative].

[00:38:41] Dr. Steven Gerali: And so the area of industry that we're in is one that is so awesome and so neat that everybody wants to get into it, and it's only gonna benefit us all. And it's like the folks that were paving the highway system decades and decades and decades and decades ago, right?

[00:38:59] Natalya: Yeah.

[00:39:00] Dr. Steven Gerali: They set up what we enjoy today. Well, what are we gonna set up for the next generation?

[00:39:04] Natalya: We don't even think about that, right? We don't even think about what we enjoy today in terms of the highway system.

[00:39:09] Dr. Steven Gerali: Right. We just, we just, it just, it's there.

[00:39:10] Natalya: We just use it.

[00:39:11] Dr. Steven Gerali: It's the same thing with the internet. Like, when I tell, talk to my kids about, "Oh, I had a dial-up and it would take forever to load a pic, "Daddy, you're silly. You're, you don't make any sense."

[00:39:19] Natalya: Yeah.

[00:39:20] Dr. Steven Gerali: They just take it for granted. And eventually we'll get to that point where space and the infrastructure will be taken for granted. But until we get that, it's something we gotta go build, right?

[00:39:31] Natalya: So you have really painted a, a remarkable picture of the future. And you also mentioned Star Trek and talking watches. Star Trek also had the ability to, you could vaporize and appear in different places. Tell us a little bit about what you envision for gadgets in the next 30 to 50 years in space.

[00:39:51] Dr. Steven Gerali: Yeah. You know, on, on the gadget side, we're gonna have to have a whole way of communicating that's completely different than what we have today. You're just starting to see glimpses of it with AR and VR and the whole metaverse, right? I know Facebook's trying to push a lot of these components and concepts and whatnot. But it's really, how do you make that experience so much more real for the end user to where it just becomes a part of them, right? It's just part of their kind of everyday li- li- living experience.

[00:40:20] And the way that I kind of look at on the gadget side is the way that my kids are interacting with iPads and AR and VR and everything else, this is just becoming intuitive to them. They're like... I can literally put my son in a VR, he's like 10 years old, and he will literally go to town. And he's amazing to watch. He can build things and create things and design things.

[00:40:44] And, and I look at that and I go, "Well, yeah." Think about it, we do it today with our customers. We have a group called our Pulsar group that effectively recreate visualizations for our customers so they can see the finished product. They can see how it's gonna run and operate. They can see how it's gonna run in mission. We can run simulations and show you every aspect on how we understand the problem and what we're gonna do to tackle it.

[00:41:08] And those are all critical things. But the workforce of tomorrow is learning those things today. My son's already 3D printing stuff. He's 10 years old. I'm like, "I, I didn't learn that in engineering." A lot of my engineer friends haven't learned that, right? We gotta go learn it 'cause they're now teaching it in, right? And so it's how do you learn, you know, the newest and, and greatest capabilities?

[00:41:32] How do you bring that to bear? How do you engage with the, the, the STEM workforce and where they want to go with it? So, you know, Lockheed does a lot of great things and I'll tell you about three of them. The first one is, what we refer to as kind of the first group. And what they do is they do Lego Leagues for, like, elementary school students. They do high school robotics competitions. They do some really neat things to really get folks interested in STEM.

[00:41:58] When we get them into that realm, we start doing, like, different competitions. So we have, we just finished CYBERQUEST here this last Saturday. And effectively, we had 250 students from around the world competing in cyber tests. And, you know, we talked about security before and we know our adversaries are always looking for the holes. So the way that we can best fill those holes is get the right professional talent that can help us pinpoint them and fill them and make sure that we're operating under our best conditions.

[00:42:27] And then we need folks that are gonna build that highway for space. Which if you think about a system today, a system is just a lot of s- other subsystems put together. Well, how, what makes that all run? Software. So what do we have? We have a thing that we call our Code Quest, which actually happens at the end of April, which is for a lot of our high school students to teach them about computer programming and to get them involved in competitions.

[00:42:53] All this is to grow that base. It's to take them from when they're little, all the way up. Because we lose so many STEM people. We lose them because either they just don't think they can do the job, or they think it might be too hard or I just don't know what it all means. And if we don't engage with that community, we're not gonna have that talent. If we don't have that talent, then infrastructure doesn't get built.

[00:43:16] It's not gonna be me building the infrastructure. It's gonna be our kids helping to build the real final infrastructure that's really gonna be the freeway of the space community, right? And so for me, it's all about... When we talk about gadgets it's how do you interact in order to solve problems and build capability. And whether it's AR/VR, whether it's digital assistance, whether it's our digital twins for how we're gonna build and operate our solutions.

[00:43:44] Whether it's any of these things. They all have importance, right? They, they walk a path to show how we're gonna actually design and engineer the future of tomorrow. And if we don't engage properly with each and every one of those students, if we lose them, we've lost our competitive edge. If, people always ask me, "Well, how does the US stay competitive?" I am telling you put your dollars in STEM.

[00:44:09] 'Cause for every engineer we build and we can go and build the greatest things, it, it has demonstrable impact on the economy, on the future of the US, on the security of our home space, on everything that we do. So all I can tell you is, gadgets wise there's lots of technologies, lots of things that we're working with students. On STEM, it's all about how do we engage with those students and get them interested and keep them interested and motivated and partner with them to grow that talent 'cause we're gonna need it to keep building the future of tomorrow.

[00:44:42] Natalya: You're talking about a handoff and a tipping point where we're... We hear at Lockheed Martin are spearheading all of these changes, but in, the next generation is the one that's gonna make those tangible and fundamental to our everyday living.

[00:44:56] Dr. Steven Gerali: Yes.

[00:44:57] Natalya: Thank you for that. I've been speaking with Dr. Steve Gerali from Lockheed Martin Space. Thanks for joining us today, Steve.

[00:45:02] Dr. Steven Gerali: Thank you very much. I appreciate it.

[00:45:04] Natalya: All too often, we take for granted the basic infrastructure that enables our daily lives. When we turn on a light switch or order a package online, we don't think about everything going on behind the scenes to make those things possible. In the same way, Lockheed Martin is pioneering unique technologies and new innovations to make the future of Space a reality.

[00:45:29] And the future of Space may not be as harmonious as we hope. As more private companies, organizations, and countries move into the ultimate high ground of Space, it will become more contested. So how do we keep America and its allies secure in the 21st Century? How do we empower the 21st-century warfighter? We tackle that in our next two episodes of Space Makers.

[00:45:55] Host: You've been listening to Steve Gerali who is a space maker. Whether you're a software engineer, systems, engineer, finance, or HR professional, we need space makers like you to make the seemingly impossible missions a reality. Please visit this episode’s show notes to learn more about what you just heard in this episode or the careers available at Lockheed Martin. If you enjoyed this show, please like and subscribe so others can find us and follow along for more out of this world stories. To learn more about our missions, products and people, follow our new Twitter handle @LMSpace and visit lockheedmartin.com/space. Join us on the next episode as we introduce you to more space makers.

[00:46:42] Space Makers is a production of Lockheed Martin Space.

[00:46:45] It's executive produced by Pavan Desai.

[00:46:48] Senior producer is Natalya Oleksik.

[00:46:50] Senior producer, writer, and host is Ben Dinsmore.

[00:46:53] Sound design and audio mastered by Julian Giraldo.

[00:46:56] Graphic design by Tim Roesch.

[00:46:58] Marketing and recruiting by Joe Portnoy, Shannon Myers, Mallory Richardson, and Stephanie Dixon.

[00:47:03] A huge thanks to all the communication professionals at Lockheed Martin who helped make these stories possible.

[00:47:10] Thanks for joining us and see you next time.