Episode #2 - Sandor Felber - Humanoids for Hazardous Operations

Today I'm sitting down with Sandor Felber, the CEO and co-founder of Minerva Humanoids. Sandor is a good friend of mine who has spent his career deeply involved in the robotics world. From his time at university building autonomous electric vehicles, to working on the Optimus team at Tesla, to his time as a researcher at MIT researching intelligence for humanoid robots. Now he's building humanoid avatars designed to enter the world's most hazardous environments so that humans don't have to. We're going to talk about what he's working on now at his startup Minerva, why the energy sector and public safety is the first frontier for their humanoids, and how we keep the world running without putting human lives at risk. Sandor, welcome to the show. So Sandor, I'd love to just start by asking, I mean, you've spent your entire career so far in, in robotics, even, I know at university, obviously you were building EVs, Optimus, MIT, you're working on humanoids, and now you've co-founded a humanoid robotics startup. What got you involved in the industry initially?

Well, first of all, thanks so much for having me. And then really what got me excited about this space is as a kid, I was super excited about anything and everything that moved and that moved fast. I think that's kind of like how a lot of like the younger generation might feel like, like, you know, when you like play with those like RC cars and like—

Yeah, like Scare Electrics and stuff like that.

Exactly. So that was a lot of fun. And I was like, how cool would it be if we had a full-size version of this? And guess what, right? At the time, Tesla was getting built as a company. And so that was somewhat influential, I would say. And then I also had a love for Formula 1 and kind of like racing and all that. That was part of it as well. We used to go to Hungaroring. Which is the Hungarian Grand Prix for F1. So I always had a kind of love for, for like engineering really high performance things and pushing the boundaries beyond what's, what's possible. Um, I, I also really love the idea of kind of getting into car racing, uh, but then I realized like that, that space is, is a bit like you have to start very early and it's, it's quite competitive. And, um, because it didn't quite feel like I have the foundations and I'm a little too old for this, like I'd rather figure out how to actually build these systems. And that's what really drew me into engineering. And that was a direction that I was definitely considering for a long period of time. It definitely got me into, you know, do my double major, electrical mechanical engineering at Edinburgh, which was a lot of fun. I really enjoyed the courses and I really liked how kind of the program was covering both fields. And also I realized like, you know, at the end of the day, most systems that we use today from the smartphone to buses to airplanes to even buildings, they're largely electromechanical systems. And so if you understand this discipline well, then that means you have, you know, a knack for pretty much everything in the world that's actually doing like work in the sense that like, you know, exerting some power output, there is some conversion of some energy form into another one, and there's some like, you know, hopefully some useful work being done.

And why, why humanoids then? You know, you're interested in cars, engineering is obviously, I mean, touches all areas of life. Like, how did you end up actually in the humanoid space? Why aren't you building F1 cars right now?

That's a great question. I— it was one of the biggest questions in my career so far, to be honest. Um, kind of getting away from, from Formula 1. A lot of my peers, a lot of the people that I've used to work with, they're now working at some of the most prestigious Formula 1 teams. And they're loving it. And for me, really, at the end of the day, it was a question of intelligence, honestly. I liked doing all the kind of electrical and mechanical engineering work and whatnot. But at the end of the day, I felt like the improvement in these systems is kind of like 5 or 10%. And really, it was at Tesla where I really realized, right? Like, if I If I want to build, say, electric actuators, which are effectively devices that convert some electrical power to mechanical kind of spinning. So if I'm to become an expert at that, then I can improve this field by a couple percent at best. And that's if I'm a really exceptional top 5 person in a group of, say, top 5 people in the world. So that's the maximum impact. Whereas I felt like there's a lot of things in robotics that are currently being revolutionized by AI. And I kind of gotten into that. Obviously my work on Tesla Optimus, which was, you know, at the time Figure wasn't even a company and it was very early days in humanoids and it wasn't intuitive. It wasn't intuitive to myself either. Like why would we use humanoids for factories? And to be honest, I am not a big believer in humanoids for factory applications except for a select few roles. But I do think that, you know, there is certain advantages to the form factor that you can leverage in certain specialized applications. But to be honest, at the highest level, what really drew me into robotics and humanoids was the excitement that like, you know, these are many, many dimensional systems, very complex, lots of kind of system designs, everything has to work really well. And it's almost like, orchestrating a very, you know, nuanced machine. And the idea of that is really what drew me into robotics and then being able to do that from a technical point of view.

Yeah, yeah, that's awesome. And obviously now you've gone on your own, your own journey. You founded Minerva. I'd love to ask just like in your own words, you know, what's the goal of Minerva Humanoids and what are you doing? And I guess, you know, a follow-up question to that is I'm sure you had the opportunity to stay at you know, in the Tesla Optimus team build humanoids? Obviously, they've got the distribution and the capital, like, why actually make the decision to go out on your own? But maybe we start with like, yeah, what is Minerva? And why did you found it?

Great question. Minerva's mission is to protect human life in the most dangerous jobs of the planet. And we're leveraging humanoid robots to do so. It turns out that some of these very dangerous jobs that we are looking at actually were around the humanoid form factor almost by coincidence and coincidentally, right? Like I'm an expert in that space. So I felt like the pieces of the puzzle kind of somehow aligned really well. I really think, Sorry, what was the other question?

No, no worry. I was going to ask why move on from the Tesla Optimus team? Why actually build something yourself?

And so I decided to build something and start something on my own because I felt like there's a lot of human-made companies out there right now that are potentially building for the wrong application and are asking the wrong type of questions. I feel like they're pursuing like all of them pretty much, the wrong paradigm, which is that, you know, they're trying to put humanoids into existing jobs and have them work around humans, which I think is not like the right first application for humanoids. The reason being that humanoids and robotics as a technology is kind of very dangerous in and of itself. And so say I were to put a humanoid in your home and in your home there were, say, kids, potentially pets, or whatever other living organisms, you would really hate if that humanoid caused any sort of damage to value property. But also, especially if, say, the humanoid robot fell on your pet, right, potentially causing some irreparable harm to you and to your pet. And so that is, that is where I think a lot of the industry is going wrong, that they don't really understand the limitations of, of safety and robotics as a, due to that complexity and due to that kind of, you have to really finesse the target application, I think, and go into a space where you don't really mind falling or you really don't want humans in those spaces. And so where do you really not want humans? It's very hazardous jobs, right? So that's how kind of, that's one paradigm that shaped I'm thinking around Minerva.

What roles are you going into initially? What are considered the kind of dangerous jobs that Minerva is going to tackle first, if you can say? Obviously, I know you're still kind of early days.

Yeah, that's a great question. So actually, you know, for us, the energy sector is a great application space. One of my, from my graduating cohort in Edinburgh, which is, you know, close to the North Sea, there's a lot of sort of oil and gas in that space. Even to just like, um, be able to fly out, he would get, uh, strapped into a mock-up helicopter into a simulation, and then you'd like, uh, be submerged in swimming pool, uh, um, capsized, so you'd be flipped upside down. Would have to literally like, uh, get out of the helicopter and prove that you can do so, because, you know, if anything happens to that helicopter out there in the North Sea, which is quite, quite rough, uh, weather conditions then you have to be able to do that because otherwise you have zero chance of survival. And even then, like, even if you can actually do this, even then you only have a couple of percents because, like, you know, the currents and, like, the water is quite cold. And so there's a few minutes to get you out of there. And, you know, just like, there was a lot of inspiration that took us, that took me from, you know, my graduating cohort and from the people I've seen doing this kind of dangerous jobs. So that's on the one side in energy, where, like, you know, there's a lot of these critical energy generation, where we still have humans in these jobs and we pay them an incredible amount of hazard pay because of the risk they undertake that they really shouldn't. And then the second one is really in public safety, where a lot of people currently are risking their lives to, to, you know, like clean up chemical biological spillages, deal with biohazardous substances. Like, for example, when it comes to, you know, potentially an active biohazardous substance sort of cleanup or decontamination, you you would have to wear these giant hazmat suits from the movies. I'm sure you're familiar with some of those, but they would often be like pressurized as well. And it's oftentimes quite heavy. Like it's really limiting your ability to work comfortably. Like it's, if anything, this is super uncomfortable and oftentimes you have to do this in the dark. So I spoke with a couple of people that have done this. I mean, one example in specific from one of our advisors was when he actually had to go, he was serving in the, in the Army and the Navy. And like one of the jobs that he was assigned on a boat was to actually go into one of these rooms that was somewhat flooded and to find some potentially dangerous equipment and look for it in the dark. I think being able to see anything in there with a single headlight that would be strapped onto his head and was like completely pitch dark in there. And he was like, you know, searching for this potentially dangerous device in like a flooded compartment of the boat. in like this overpressurized suit. And if, you know, anything happened to the suit, there was also some chemical or biological risks to him. So it's, uh, uh, the risks these people undertake are quite like large, and I think they're definitely excessive to, you know, what they should be undertaking. And, and, you know, unfortunately right now the doctrine is that we just consider these people expendable, um, in some of these missions, uh, for, you know, the greater utility of society. But at the end of the day, I think like technology has a great role in pushing the boundary for this kind of higher and for people to not be doing these jobs, only a select kind of few of these jobs that our latest and newest technology will still not be able to perform.

Yeah, earlier, Sandor, you said that you weren't actually too optimistic about humanoids necessarily as the kind of optimal form factor for like in the home or manufacturing. We'd love to ask like why for the use cases that you're tackling have you decided to to opt for humanoids rather than custom builds for the specific problems you're tackling?

That's a great question. And I think, you know, you also almost answered your question in and of itself because the humanoid is the custom form factor. The reason I'm not so bullish on these other general purpose applications, and for the most part, the narrative that I hear when it comes to humanoids is that you have some general purpose application and the humanoid can do all of these things because it's a humanoid, it can do everything a human can in all these different environments. But I, I've, I always felt like that's not really quite true. So if you look at, uh, an electrician and then you look at a plumber and then you look at say a, um, nuclear power plant operator, their skillset, their training, their understanding of the world will, will be almost entirely different. And even the, the, the personal protective equipment, the PPE they would wear would also differ quite a bit. And the tools they would carry on themselves would also differ quite a bit. And so at the end of the day, I think we, you know, we have specialized humans to be doing these jobs, right? So that's, that's hence, you know, we have these fields and professions. And the way I think about this is that the humanoid is actually the ultimate specialized custom form factor. So we are actually customizing quite a lot on our own humanoids. What we are building is nothing like what you would see from, the other sort of big brands that are out there today, because we are really customizing pretty much everything on our humanoid for these target applications that we're tackling. And I could go through every single part of the robot, but the basic mechanical design can be somewhat similar to other humanoids. But at the end of the day, a lot of the equipment on the robot, a lot of the kind of electrical and mechanical design and capabilities of the humanoid that we are building are custom tailored for these applications, which require sort of a higher force, lower dexterity levels than in a lot of these other cases that they're going for. So for example, in the home, you really need very delicate manipulation. For a lot of the use cases that we're tackling, that's not true. So, you know, for example, that's one trade-off that we are taking there. Another one regards the battery life. For example, for us, we don't have to go for that long, so we don't need a huge battery. Whereas, for example, a home robot, you might want to have a longer battery life. Let's say less robustness to, for example, falling. We do customize a lot of our humanoids to the target environments. But also, if you just think about for a second, like, who these kind of environments have been designed around, right? Like, they have pretty much been designed around the average 5'9" tall adult population. Is, you know, around humans. Every single valve, every single gauge, every single kind of operational equipment that's been designed in these environments has been designed around humanoids or the humanoid form factor as a whole. So, you know, if you want to exert a lot of force, you would find that it's quite tricky to have that, to exert that amount of force if you don't have the purchase that you as a human can have when the valve is conveniently mounted right in front of you and you can turn it. Or for example, in an electrical or nuclear power plant, right? Like all of the gauges are at human height. So for example, if we compare our form factor to quadruped dog-like robots, which I wrote my master's thesis on quadruped robots and I'm a fan of them. I like that they're amazing, but I do think, you know, they're not very good for applications that are at a sort of a table height above because you just don't have the vantage point you don't really have a camera high up that's kind of intuitive for a human to use. And this is the last bit, intuitiveness. At the end of the day, a lot of these jobs are about just like being able to perform them in a reasonable amount of time. And if you have played with robotics, which I mean, I'm sure you have, but maybe some of the listeners haven't, then effectively people that are deploying robotics in real-world field applications, they all know that it's very hard to do very basic tasks that sound or seem basic to us as humans with robots. And there, you know, it's one thing maybe interesting to consider is Moravec's paradox, which is that what seems easy to us is hard for computers and machines, but then what seems hard for us is easy for them, right? Like it's much like the calculator, like it gives you about like, by, you know, 1,025.8 over 58, 59.5 is like quite quickly, quicker than, you know, probably any of us could. But at the end of the day, can it actually, you know, do some generalization? Can it actually like move in the real world? And no, it cannot. And, you know, a lot of the kind of capabilities that we have in LLMs, they don't— those capabilities don't just transfer to robots. Those capabilities have been trained of data that was widely available using the internet. And that's just not true for, I would say, probably like 80-90% of robotics. And so, you know, there's a whole data collection problem around this as well. Effectively, what that means is that, you know, you can't just assume that AI is working for all of this and can help you kind of transfer your intention into a complete task, right? Like in a lot of these cases, you intend to say, you know, move the lid or like say, unscrew a bottle. You cannot just transfer that intention that easily because we first of all need to collect the data and we need to have some intuitive devices to do that in the first place. So we are taking a very sort of, um, first principles, ground-up approach here, and we are building that architecture, um, assuming that, you know, the AI capabilities are still not going to be mature for the next few years at least.

Yeah, yeah, it's, uh, it's, it's, it's really interesting. And I completely understand, like, okay, so the world, the world today has been built around, like, humans doing these jobs, right? So it makes sense. So the example you gave of at the nuclear power plant, the wheel or whatever it is, is it like human, human height? So humanoid form factor makes sense to deploy. And with certain levels of like dexterity or like robotic abilities, you can actually complete a task. I think you've seen in certain environments that say like warehouses, that actually the environment has started to conform more to a robot than to a human, like companies like Ocado or like Amazon, for example, have these like fully autonomous warehouses, like grid systems. Do you see longevity with humanoids in the places that you're deploying? Or do you think over time as robotics advances, actually, you'll start to see like the change to the environment rather than necessarily like keep— yeah, and therefore maybe like changes to humanoids as the optimal form factor for solving that problem?

I mean, I don't see the reason why we couldn't build a fully automated, you know, end-to-end automated say, oil rig or power plant for electrical, like hydro generation, right? Like it could be totally autonomous and with like a special set of robots that are mobile robots that walk around and a number of sort of robotic solutions that are built into the system in and of itself, right? I think what we have to understand is also technology has sort of a, it's all, there's always a right place and right time for technology, right? Like If the steam locomotive would, would, you know, try and make a comeback today, it would have a very hard time to do that because the energy efficiency is just not comparable to what we have access to. In the, in the 19th century, it was revolutionary, started the Industrial Revolution, right? And so in, in my view, that is still very far away given that, you know, um, power plants are getting designed in, you know, a cycle of like oftentimes decades, and we don't have any expertise designing them around custom robots. And we don't know how robots, like what's easy for robots even. We humans, humanity sort of as a whole, we don't have a general good understanding of how robots behave in these environments. Like really with humanoids, right? In the 2010s, there was a couple of research projects around humanoids, mostly in for space applications, but that was it pretty much. For quadrupeds, we have seen some more successful deployments and kind of like some more industrial applications and some penetration into some markets, but, um, it's also sort of just picking up. So I would say we don't at all have a good handle on like what these robots can do really well and what they fail at and what's the best form factor for what. So there's sort of a fundamental design question around this and, and you know, frankly, if any of these robots go down or anything happens to any of these systems, then you still want to be able to fall back onto something that's more reliable. And if, say, you cannot reach it as an average human, because say your robot is a bit taller than a human and it has a better reach, then that's a problem, right? Because then you suddenly can't do it with a human. So you can't just like replace. So I think there is something about that replaceability. That really is helpful in sort of market adoption. But also it's kind of a paradigm question, right? Like, do you, do you design or potentially fully retrofit an existing, you know, energy generation plant from the 1970s? Or do you just like deploy a couple of these robots and use them? And I think there is some cases where one of the options makes sense, and there's some cases where the other one will.

So Sandor, maybe going back to something that you, you said earlier, you know, you're custom building your own humanoids. And I know you probably can't get too much into the task that actually you're looking to tackle, but there's a lot of really great hardware existing, like humanoid hardware out there. You know, it's like Unitree, Booster in China, there's US players, there's European players. Why actually go through the effort of like R&D and manufacturing your own humanoids for your use cases?

So firstly, I think people I don't really expect that humanoids will appear in very different shapes and forms in the future, but I personally anticipate that most humanoids will not be as human-like as we think today they will be. So a lot of the companies, I think they're just striving to make as much of a human-like robot and they're making an active branding and kind of design effort in their humanoid, like designing the covers, their, you know, the colors they use, the materials they use. The shapes they end up putting on their kind of mechanical structures that then makes them look like they're just another sort of human. So they're effectively trying to build these proto-humans. But at the end of the day, I think this pace is going to be much more about the humanoids that won't look like humans exactly. And there's like little quirks around them that will make them superior to humans in a few ways. For those specific target applications. And again, like, I think it all boils down for us, it's really a specialized application that we're taking a specialist form factor in, which just happens to be a humanoid. So, you know, if you look at, again, like one of these like application spaces, which, you know, it's got stairs, it's got handrails, it's got valves, gauges, levers that are all at human height. Whenever you're trying to, you know, say carry a heavy object and, you know, hold onto the handrail, it would be quite hard to do that with, say, a quadruped that has an arm protruding from the top. So one of these dog-like robots with, you know, with like two legs, two back and two front, and then one arm protruding from the top. That would not be the right kind of form factor to be doing that. However, I do think that, you know, for example, some applications of, say, having an extra, say, pouch in the front or the back of the robot, right? Like, like a lot of us walk around with backpacks. Why don't we just integrate that into humanoid, right? So that's, for example, a thought that I've been having for a long time. And, and, um, I think there's a lot of very exciting use cases and a lot of very exciting customization opportunities that people should try and discover more aggressively in the industry. And, and I'd like to kind of give that push to, to everybody in the space that, like, there's, there's some really good ideas, I think, out there that we still are just completely, um, unaware of. And so let's go out and explore. I think it's a very exciting time to do that.

Wait, when, um, when will we be able to see a Minerva humanoid? Do you have a timeline or, uh, kind of TBD at this point in time?

So we're hoping to have our first prototypes before the end of the year. Let's see where that goes. What we're doing is really custom, and we're using a specialist supply chain as well for our own humanoid for a number of reasons. And a lot of the stuff that we're putting into our robot is, there's little like one company that's capable of making it. So I think contrary to a lot of others that think commodity is hardware has been commoditized in robotics, especially in humanoid robotics. I think, you know, given that the space has existed for, you know, it's a nascent industry, it's existed for maybe a couple of years at best, but it's still, I wouldn't say this is an established industry still. And the growth of the industry is incredible, but most of the players are still kind of, you know, R&D stages and having some prototypes that kind of work, but not really. They support, say, one programming language, but not the other. They support, say, be, you know, one type of operating system but not the other. And there's a lot of just like trial and error that goes on, um, by everyone today, um, not just on the AI front of like, you know, how should we collect data? Should it be just vision only? Should we actually consider other sort of multimodal data collection methods? Or should we just really try and get vision to be as good as possible? And is that enough? So I think as much as that goes on in the AI world, it also is going on in the hardware world. But I do think there's a bit less hype about how that goes on in the hardware world. At the same time, I do think it's kind of just as exciting that the hardware is kind of like, you know, for the first time maybe in a new industry, now we have AI and that kind of is changing how we are designing these robots, right? Because we can design around the kind of these policies that we train, these AI models that we train, these foundational models that we put on our robots. and we can get the robot's brain to actually understand the limitations of the hardware and also leverage the unique strengths. So if you're able to, as I mentioned, like put in some kind of unconventional, kind of crazy looking quirks into these robots to do some tasks extremely well or efficiently, then that could be, you know, a great way to have some specialized humanoids, you know, even as, you know, like you could think of something as, mundane as being just like, you know, the cashier's desk, right? We're picking up objects with great speed and just like scanning them through, uh, probably will run something a little more dexterous than just like these kind of currently existing hands, which are kind of tend to be slow and have, you know, not, not great robustness to say the least. And, um, you know, you could probably design some sort of a gripper that's capable of gripping most packaged objects that are, you know, unopened. To do this job really well. So that's, you know, for example, one of the customizations that I could imagine for that specific role.

Yeah. And I guess on the note of like AI and like policy training, one thing that I thought was interesting on your website, you know, typically you go on these like humanoid or like AV companies websites and it's like autonomous robots deployed today, etc. And you know, what I thought was interesting on the Minerva Humanoids website was, you know, you speak about actually like remote operators. Would love to just ask, like, yeah, what's your view for autonomy with Minerva? Is it like fully teleoperated? Is the goal to become fully autonomous? How do you think about autonomy with, with your application specifically?

We're going to have fully autonomous humanoids eventually that I think will be able to communicate with each other. You know, much like you have swarm drones today, I think that's going to be You know, given, given enough time. Now it's very hard to predict where the time is, and I don't really want to make any sort of predictions, except for the one that it's not going to be today or in the near future, because we're just not there. Tools are not there. We don't have the data. We don't have much of the puzzle, to be honest, right, to even start with. So we need to first, you know, kind of gather the pieces around so that we can kind of try and put them together and see how they fit. So I think to summarize, that's where the field is right now. And so anticipating that, we rely a lot on sort of intuitive operation by humans. And I think also a lot of where a lot of these companies go wrong from kind of a business and economics point of view is that if you're doing something that's sort of a wage arbitrage thing where you are like, there's some worker that would perform that job for X amount of dollars per hour, and then your robots can do it for, say, $2 an hour cheaper. And, you know, in the limited sort of kind of switching. But then there's also a lot of kind of anxiety and kind of like worry about like, okay, well, right, like, if I do this and I replace, say, the human with the robot, and I say I need to maintain the robot every month, then that might just, you know, achieve my profits. And it actually even might be more expensive to use the robot. And if the robot can't resolve a couple of things and you still need, say, one person to go around and even, you know, say once a day, try and like make sure that, you know, the one task that the robot couldn't like completely somehow perform, like to make that happen and finish that task, then that will really be a big issue for a lot of these companies. And I think in that way, I'm not so bullish because the unit economics is hard to make it work in a competitive economic landscape with, robotics, which traditionally was reserved for really high, like sort of highly specialized, like much needed spaces, right? Like if we consider where robotics has more success, it's pretty much been aerospace, some nuclear automation sort of things and sort of like airports, but there's like, you know, enough numbers. So I think it's either like some crazy high number of things where you can custom design a whole system and whole environment around the job. You know, the Amazon warehouse, I think is a good example. Or it is really some specialized low-volume application, which is like space. So I think it's, you know, but one interesting thing that I think, you know, Tesla is pursuing now, which is space with Optimus. And I definitely like, like think that that application is very interesting. I think I like the approach that they are taking that like, hey, let's, you know, try and send humanoids to Mars and see if we can colonize it. I do think it's an ambitious mission, but if you think about like, we as humanity have expertise to build spaceships that we have developed for the past 50, 60, 70 years around humans. And if I asked an aerospace engineer to design a spaceship around a dog-like quadruped robot with one arm protruding from the top, they would not have much intuition about like how this could behave potentially, right? They wouldn't really understand like what are the constraints with this? What are possible failure modes? There's a whole lot of testing that's required there. Like, we get to skip a whole lot of that because if you, if you do have a humanoid and it behaves kind of similarly to a human and you can assume that it can kind of press buttons in the same way and, you know, operate machines in a similar fashion, then, then you do get to skip all that kind of redesign, if that makes sense. And that's why I'm bullish on humanoids in a couple of these different kind of unique applications. And I think there's definitely spaces where they do make sense. But I just don't really buy the general purpose story.

Yeah. Yeah. Interesting. I'd love to, I've got one more Minerva question to finish us off, but maybe just a penultimate question. So I'd love to ask then, like, I mean, do you see the companies that are looking to deploy humanoids in the home? I won't name any names, but like, I don't know, like, do you, there's a lot of companies that are trying to be quite ambitious with their rollouts there. Um, do you see a world where we do have humanoids in, in the home, or do you think that's actually further away than, than most people think?

I mean, I think there is some really cool, especially smaller form factor robots that are, you know, they're like lighter, they're smaller, they're like kids. And I think I'd be a lot more comfortable personally to have a much smaller robot than a full-size humanoid in my home, you know, running around doing anything, to be honest. Both from a sort of a personal, like, anxiety kind of security perspective. I mean, you know, people do get quite paranoid and they don't tend to be quite open about this, but a lot of, a lot of people are quite concerned about, like, privacy, security, all that. And so having these kind of highly intelligent devices that are collecting data as they're, you know, out there, because a lot of these companies are, you know, actively trying to do that, people are kind of aware of that. And I think For example, for the home argument, you have to really finesse that one and also make a product that really looks kind of something that's kind of cozy, homely, fits into your home really well. And to be honest, they kind of forgot that Minerva is difficult designing. It's definitely not one that would like fit like just like the general idea of my home, for example, that, right? Like you kind of want, say, plants in your home. You want, you know, cozy blankets. You want to have probably, you know, a nice couch or something. You don't really want to have like, cutting-edge, like, engineered devices. Like, you know, imagine like kind of having sort of like car parts in your home, right? Like, a lot of people assimilate humanoids and robotics as a whole to the automotive industry and to, you know, a lot of the AI that we have for robots is actually coming from the self-driving, the autonomous driving push. And so I think, you know, those comparisons make sense. But also, when we actually do deploy them in the home, we have to think a lot about like the social circumstances under which you're doing that and whether that's the right product. But I do think they are not too far. I think we all see it on the horizon, but I think estimating the distance as it is tricky at sea, estimating the distance, you know, how far you are from land is also similarly tricky for the space. I think there's some countries as well where adoption is going to be a lot quicker than in some others because just people are a lot more used to. And then there's also, you know, sort of a smart home integration question as well, right? And then how well we can actually tackle those first useful use cases and really the reasons why you'd have robots. But to be honest, just to summarize, I also do think, you know, beyond these social and kind of like almost like vibe reasons that you want to have a robot in your home or not. I do think a lot of people will kind of like go for some weird use cases, which again, we don't even think about, right? Like having, say, you know, a companion to your elderly and, you know, watching out for them and potentially, you know, calling you in case of an emergency or somebody else, say the ambulance, right? That could be a really good application, for example, where I do see you know, humanoids or other types of robots emerging quite quickly. Elderly care, I think, is a great application for the home. So there's definitely like, like interesting, very, very bullish on some of these cases.

Final question, just to finish this off. Minerva, if everything goes, goes to plan, what, what does the world look like with Minerva in 5 to 10 years? What are you doing? Are you still focusing exclusively on kind of like public safety and energy? Are you still just humanoids? What do you see the future looking like for you guys?

I do think we will largely stick to a similar form factor to what we have right now. There might be different lines of product and more modifications around what we have currently. But at the end of the day, I think it's going to be more about the autonomy that's going to improve and sort of the, you know, the cost-benefit analysis is going to look a whole lot more obvious and a whole lot more kind of appealing to the average Joe as well. But it comes to these use cases, we really want to, you know, put people out of harm's way. I like to say we measure success in the number of lives saved and the number of robots blown up over other factors that I think a lot of companies are measuring their success in. I think even if we, you know, don't end up deploying robotics at a mass scale, I hope that we can actually usher in one of these good applications for humanoids that hopefully will be successful. And I'm very bullish on, you know, our use cases because to be honest, human lives are at stake. And, you know, I do have a core belief that human life is valuable. And we as a society, we also, you know, in the UK, in the US, we also do value this a lot. And I do think if we are able to do life-saving work, then that means we're inherently doing something that's valuable, and that will be driving value. And hopefully we can save lots of lives as we undergo and figure out some really good customizations and specializations around the use case as well.

Well, Sandor, that, that's awesome. I'm obviously like super bullish on what you guys are building. I can't wait to see the humanoid that you come out with, hopefully before the end of the year. And Yeah, I really appreciate you giving me the time. So thank you so much.

Yeah, very excited as well. Thank you so much, James.

Perfect. See ya.

See ya.