speaker-0: Hello, welcome to the Robotics Tech Podcast, episode number 11. Today we got a very special one for you because this is our first guest. His name is Bryce Chanciotto and he's working on an X radar. So we're gonna go very deep on this topic. My name is Oscar. ⁓

speaker-1: My name is Leo. So we are with Bryce Chanciotto today. He's designing an actuator that behaves like a human joint. How are you doing today, Bryce?

speaker-2: I'm doing great. It's it's great to be on call with you guys.

speaker-1: Yeah, it's great to have you. We are super excited to learn more about it. Can you tell us a little bit about your background first?

speaker-2: Sure. So right now I am a student attending a university called Cetasaw State, based out of Georgia. ⁓ outside of my studies, I have started a few startups. ⁓ one of them was around a kind of autonomous car kit that you could install in your car, allowing it to drive by itself, kind of like Kama AI. ⁓ the other one was this exoskeleton you could wear that could provide resistance training. ⁓ long story short. I'm no longer doing those and I'm, you know, kinda working on this one right now where I'm developing the orbit actuator. And I've been doing this for quite a while now. The very first ⁓ iteration of this tech was developed back in twenty twenty one. ⁓ and it just kind of been chugging since. I'm

speaker-1: I'm gonna dive a little deeper into iTraders fundamentals before we get to your product. So if I had to summarize it, I would say that iTraders are the muscles of a robot, right? Would you agree with that? Absolutely. Good. So we are on a good start here. Okay. So modern electric robot motors, they can hit three to seven kilowatts per kilogram, while a human muscle can only hit zero point three kilowatts. So in a nutshell, that means that a robot would outstrain you by several orders of magnitude. So don't get into a fight against a robot. An actuator is an essential part of a robot because it is a component that converts stored energy into four stalk or displacement. You could frame it as a robot needs sensors to perceive the world, and then it needs actuators to interact with the world. I like to present actuators in like three levels of classification. First of them would be the type. ⁓ so we have hydraulic eye trailers who push something with pressurized oil, pressurized water. They are by far the most powerful, they are usually pretty large, but not really precise. In between you get pneumatic actual. It's the same pushing mechanism except you do it with air instead of a liquid, like oil. It's mostly used in industry and automotive right now. It's less powerful but more precise and hydraulic. And then we get electric. They are extremely efficient and they became basically the standard right now. Robotics, at least for you, by the way. I'm I'm not quite sure about the rest. So the second level of classification would be motion output. So you get rotary that delivers torques and linear that delivers force. Do you want like rotary force or do you want linear force basically? And third classification is transmission architecture. So you get the harmonic drive, which is more precise, more stiff, more torque-dense, but is not backdrivable. And then you also have quasi-direct drive, which has a larger diameter and it's mostly used in legs. If I had to summarize it in a nutshell. Of course there are exceptions. The way I understand your product, Bryce, is that it's a new level three architecture, so it's a new transmission architecture, but with like a level two promise. Would it would you say it's correct? I'd nice, nice. So can you tell us a little bit more about your new type of actual drive? Yeah.

speaker-2: That's accurate. ⁓ So ⁓ I dubbed it the orbit actuator. And how I view it is ⁓ you know, you're referring to these things as like a level two. ⁓ I call them primitives, you know, linear and rotary, they're both primitives. They both kind of serve as the building blocks for pretty much every robotic system if they want to move. But, you know, if they want to recreate ⁓ these robots, if you're trying to recreate the motion that you see like a human body, but at the shoulder or the hips or the wrists and the ankle, that's not single axis motion. It's three axes, right? You've got pitch, yaw, and roll. Now, what engineers typically do is they'll take the single axis ⁓ actuators that you have, like an electric motor typically, and they'll stack them on top of each other. So you'll take a single axis that'll represent the pitch, and then you'll put another actuator on a different plane. This one will represent the yaw and it will spin independent of that first motor. And then you'll take a third motor and stack it on top of the second motor, and that'll be the roll. And there's other ways to do it. Mostly they use like linear actuators and tandem with heim joints or universal joints. But there's a lot of issues that come with those designs, primarily to do with like distal mass. Because most of these systems stack it's called a serial stack of actuators. Stack these actuators in such a way you have to support the weight of each consecutive motor down this chain. Doing so really reduces the efficiency of these systems. It adds a lot of problems in terms of wiring. ⁓ kind of limits the range of motion. It's harder to cool these because, you know, every single motor is moving relative to every other motor. And they're really kind of complicated to construct, especially in humanoids when, you know, humanoids are basically an assembly of like a dozen or so of these stacks of actuators. And all that being said, you know, I thought it would be a good idea to create an actuator that instead of relying on the status quo or what engineers usually rely on, to create these multi axis joints to to make an actuator that actually functions like, you know, your hips or your shoulders, or to create one that moves spherically, like a ball joint. And that's that's exactly what this actuator does. Yeah, no, I've got a bunch of different ones that I can show off later. Hopefully that answers your question.

speaker-1: Yeah, definitely.

speaker-0: So you're kind of merging all these steps into one, basically, right?

speaker-2: Pretty much. We've taken what would normally require ⁓ tens of components and removed all the inefficiencies that comes with that and just kind of bundled it up into a single package that's easy to integrate into any sort of robotic system.

speaker-0: Yeah, and ⁓ just yesterday I saw an interview with the founder of OneX and he made a good argument. He said we want to be as close to the human as possible with everything that we have on our robots because eventually it will make it much easier. He was talking about skin resistance and all these like minor things, but he ⁓ he somehow made the argument that only this will ultimately also allow them to train on video because video is usually like humans, how they actually work. kind of, you know, and move in space and stuff like that. I don't know what he was actually going for there, but ⁓ it kind of reminded me of that because ⁓ sometimes, you know, technology is making it more complicated or diverting a lot from the actual human, right? To some degree. It's necessary. Until someone comes up like you and thinks like, hey, maybe we can do it. So I think that's cool.

speaker-2: Yeah, thank you. ⁓ no, I'm I'm I'm kind of in his boat, I think. I think if you're going to try and adopt this biomimatic mindset to develop these humanoids or these robot dogs, you should be using components that address these kind of developed systems that you see in robotics and try to like optimize them to be more like the systems you're trying to mimic, right? So I totally get his point and I totally agree.

speaker-0: Yeah. What's like your take when it comes to the competition, how they are doing it and how how you see this this actuator actually coming into into reality maybe?

speaker-2: Yeah. So I think there's like two sides in terms of competition. ⁓ like we've kind of discussed already, this status quo, which is like stacking actuators on top of each other to create all the multiple axes that you need in order to mimic the motion that you see in biological systems. And then the other side are ⁓ there's a lot of novel ⁓ designs in academia that try and create this motion. I think one of them is called the Ebenex joint. That's I don't know if you guys have seen it. I know it has like millions of views on YouTube, but It's a spherical gear that kind of rotates ⁓ like this, but it's it's limited in its range of motion by this socket that the ball sits in. And it's also requires four motors to actuate three degrees of freedom. And the torque is also significantly limited just strictly because of the geometry of the gear teeth. Just to clarify, the question was like how do we integrate into the the market in reference to the current adopted technology?

speaker-0: Yeah, I mean just like where would you see it? Because sometimes you have humanoids, for example, that ⁓ are pretty open, right? You can almost modularly adapt them and maybe I'll I'll put my actuator technology on top of it, or have you already implemented it in the humanoid? I just want to grasp like where is this like standing.

speaker-2: So there's quite a lot of places that I could see it being integrated. Because it functions like a ball joint, you could see it being used in any kind of joint that's trying to mimic the spherical geometry you might see in, you know, like a person. Shoulders is a big one, wrists, ⁓ hips, ankles, even necks and like the waist, you know, waist is important. One, so you could bend over, but two, so you can kind of move your center of gravity around the humanoid. So having those additional degrees of freedom is really helpful for something like that, when especially when These humanoids are going to be used primarily in factories, lifting things up. You gotta adjust that center of gravity to get like better grips on these things. ⁓ outside of humanoids, I think it would be incredibly useful in things like robotic arms. Because the same issues you run into with humanoids in terms of these serial stacks of actuators, you run into in robotic arms. So you could reduce the the distal mass of all these systems by using something like this. You know, you could have the three axes of motion without having the weight any of the motors that you'd see normally. And then there's lots of other ones you could see being used in like gimbal-like technology, a lot of defense tech, ⁓ drone tech. One of the cool examples that I think would be really cool is in prosthetics. In prosthetics, you're not gonna see anything that really moves above one axis, specifically for like a wrist. In fact, what you'll see a majority of the time is a wrist that won't even move. You'll see a hand that's kind of just affixed to a forearm or the residual limb. So if you could design a prosthetic around this actuator that kind of adopts the same form factor as you'd see like an upper forearm and a wrist already, ⁓ you would enable these people to have all three degrees of freedom that they lost, enabling them to kind of regain the functionality that they lost. ⁓ and I think that would be really cool. And there's plenty of others I could keep naming.

speaker-0: Okay. Maybe another question that touches on that. Are some people already using similar technology or where you see this already or is no one else using this or implementing this in any shape so far?

speaker-2: So the closest I've seen in industry is this joint that has a very similar name to mine and I was not aware of the name probably. It's called the Orbita. Originally it was created by I think it was Mark Rosheim or Rosenheim or whatever his name is, but it was adopted by Pollen Robotics. And then they were bought by Hugging Face, I think is the name of the company. and they used I I think it's called a parallel spherical actuator wherein they have all these linkages that have kind of this spherical geometry and they come up and meet like a at a platform, kind of like a steward platform, but way more fancy and way harder to manufacture. But it does have this kind of motion, but it's really limited in its range of motion and the torque that it can produce. So my actuator has a hemispherical work envelope. So anywhere within a hemisphere it can move. But with the orbit ⁓ it's limited to kind of a conical work envelope. So anything past, let's say like Forty five degrees. It just can't move past that threshold. That's not much. I I think it really kind of limits any adoption that it's had. And I think the the last one that's vaguely akin to this is used in the Da Vinci robot, ⁓ for their little tiny baby hands that they use to do the surgery. They've got a really, really wicked cool wrist mechanism that relies on three or four universal joints with threaded rods. attached at the base of them and you can vary the length of these threaded rods to vary the angle of these universal joints to kind of create this motion. And it's really it's a stiff design, but you know, there's a reason you only see it in very, very small scale mechanisms.

speaker-0: Yeah. And with the Da Vinci I think Leo and just in one of our last episodes talked about it how they actually reproduced some of the capabilities of it with a unitree and tele operation. ⁓ yeah. Gallbladder, gallbladder removal.

speaker-1: Yeah, so in a nutshell it's like, you know, the the Da Vinci robot, it has like the robot that operates on the patient on one side, and on the other side there is a teleoperation device, which is like an entire booth. The surgeon is almost getting in inside of, you know. And they have control with the hand, with the feet, and they have an entire like field of view that takes their entire head for like an immersive view of what they are doing. They get this teleoperation controller and they use two unitary robots. to operate on a pick instead of using the Da Vinci. And it worked. And it's pretty impressive. Yeah. Because the the difference in pricing is several order of magnitudes as well. I think a Da Vinci machine is close to two million dollars. And the two unitary ⁓ humanoids were like 15 or 20k each.

speaker-2: I I mean I could see that it'd be being adopted across the medical industry. I don't know if Unitry would be used for that anymore though, because of the FCC thing that just happened the other day. Unfortunately, ⁓ that's a whole other topic. I don't know if we would have dive into that or not. We are pretty

speaker-0: What is your take on this recent ban? So I'm

speaker-2: bit conflicted. Like I I understand the argument from the security perspective. But I also think so it's almost like anti-capitalists in a way. I think you're getting rid of competition just to support the humanoid companies that are in America. Which I understand. If you want to be able to maintain ⁓ the industry that's kind of developing in America since it's a little less developed than what's in China right now because China is supporting a lot of the companies that are developing these humanoids. I don't know. I don't love it and I don't hate it. It has its upsides and downsides. It's probably the least direct answer I could give, I think.

speaker-1: Yeah.

speaker-0: Yeah. What do you guys think? No. I I agree. I think it's very new, so we will see how it play out, you know, all these market dynamics that come into place. It's not like you can anticipate everything. It's very similar to the GPU thing where they are not sending GPUs to China, kinda, but this time it's kinda like the other way around a bit. And there's always these other layers as well where it's okay, you're blocking the humanoid, but you blocking also this type of component. Maybe you need this one from China. This one comes here. Maybe there's gonna be open models coming there, so I think it will be very interesting how broad this will eventually get. And hopefully not too broad because eventually I think I mean, we're just better off ⁓ siding all together and not getting over ⁓ taken by some some some weird AI, right?

speaker-2: Yeah.

speaker-1: Yeah. Yeah. I personally think competition is for losers until someone outcompetes you. And I think that's kind of what's happening, unfortunately. So I I think it is I mean, I understand the move, ⁓ even though you I'm not American, so it's easy for me to say. But I also think it's a bit sad because I think research will suffer the most from it, definitely. ⁓ because there is no equivalent of unitary in the US quad, right? You cannot get a humanoid that's Good enough for 20k, it just doesn't exist. I've seen posts about OneX who wants to develop a humanoid platform for developers.

speaker-0: One X is like twenty K, I just looked it up, right? So it's kinda competitive.

speaker-1: And there's also something else because I mean ⁓ I don't remember the name of the guy or the startup, but I s I've seen a guy who has a startup in the US who does a robot OS kind of that is a hundred percent safe ⁓ when it comes to security matters. And he says that if you run your unitary robot with his OS, you are 100% sure there is no Chinese backdoor still active in the robot. So if there are options like this to run a Chinese robot safely on US soil, then it becomes a hundred percent an you know, anti competition move, not a security risk. So security risk is just there to wrap it up a little bit nicely. So yeah. I don't I don't know what you think of it. I think since it's mostly cheap robots, ⁓ the people who will suffer from it are the ones who actually need it the most. So research, academia and so on, unfortunately.

speaker-0: But maybe back to your actuator. But yeah, again in a similar sense, I mean we talked about some of your advantages. Maybe there are more that you haven't mentioned yet. And also maybe limitations where you see, okay, it's a great tech that I developed ⁓ with my heart. But maybe there's also some things where you would say, Okay, maybe maybe this is where I can make it better or some

speaker-2: Obviously, contupt to the advantages, you've got that lower distal mass. So the way that the actual wear works is you have this joint off top that's for the most part hollow, right? It's really just a shell. ⁓ you kind of see through it, it's got up through a hole. ⁓ but all of the motion that happens up here is a result of what's happening down here. So what we have here are three electric motors. But instead of being serially stacked, these motors are stacked in parallel. So they all rest concentrically. They don't move relative to one another. They stay fixed. And each of these motors corresponds with a different axis of motion up top. So this top motor controls this, what I call the lower hemisphere. Some people call it the clam shell because it kind of looks like the seashell. Then the second motor rotates the upper hemisphere. Like this, and the third motor right here rotates the end effect like self. ⁓ so you know, you get the benefit of all the motors are down here, you don't have to worry about that weight there. Additionally, we have this through hole. And what the through hole allows you to do is route wires through the center of the actuator without sacrificing the range of motion of the system. So in a lot of humanoids or in a lot of robotic systems, you have to route wires to each of the motors or like down this chain, and those wires can limit the range of motion and ⁓ Because of the way that this through hole works, ⁓ the wires go straight to the origin point of this joint. So you don't need to worry about them getting tangled or anything like that. They just move ⁓ freely as this thing rotates around. ⁓ because it has this lower distal mass, you also have like lower reflected inertia. I think it also has like better packaging. It's pretty modular, easy to integrate. ⁓ some things that I have yet to really find out is one, how well does it work in Cinder Reel? I think, and this is based on conversations I've had with this guy named Anamash Garg. He helped develop ⁓ Isaac Lab back when he worked at NVIDIA. ⁓ he thinks it would be great, but you know, that's just word of mouth. We'd have to really test it. ⁓ so I hope he's right. ⁓ but we'll see. Additionally, and I think I already mentioned this, takes what would normally take 10, 20, 30 components, mushes it all into one thing. You can imagine how much easier it would be on the engineers designing these robots to just buy a single actuator that does all the work. Of these three actu the three actuators normally that would have to be attached with brackets to one another. You gotta deal with all the bearings and all the electronics and all the wires and it's just kind of a nightmare. And this one's a little more niche, but in in those typical systems, what you run into a lot of times, and this is why Boston Dynamics redesigned Atlas again, is pinch points. This does not have a single pinch point in the design. So you wouldn't have to worry about ⁓ any of those. So it's really, really nice when interacting with with humans. You know, humanoids are kind of designed to be able to interact with humans. So you want to reduce as much danger as possible. in terms of sealing the joint, ⁓ normally it's kind of hard to like weatherproof a lot of these multi-axis systems. ⁓ the way the joint design, it's very easy to do that. It comes with lots of different configurations as well. And I'll I'll get to the downsides in a minute. I just got to go through the benefits first. So we've got the the three axis one here, obviously it's got the pitch yaw and a roll. ⁓ and then ⁓ we've also got this two-axis version. I created this as a consequence of a conversation I had with Jonathan Hurst. He's one of the founders of Agility Robotics. ⁓ he was looking for a better way to design the wrist in his robot. And I showed him the three-axis version, and he said, That's really cool, but we already have a third axis or the pitch kind of embedded in the forearm of the robot. ⁓ we don't need that extra axis, it would just be redundant. So that's what this one is. Well, so it it has just two axes of motion. Well, not this one in particular. The one I'm referring to, ⁓ the lower hemisphere's fixed and it has an end effector in the upper hemisphere that spins, but that's besides the point. And then this one is a four axis version of the actuator. ⁓ it has the pitch yaw unroll that you normally see, ⁓ but it also has two independently ⁓ rotating end effectors. So what's cool about this is in addition to having those three axes you you normally see. Say you want to stick an end effector on the end of this. ⁓ instead of having to take a motor, stick it on the end, you know, you gotta deal with the additional mass that comes with that motor that would be used to actuate the end effector. ⁓ you can use the motor that's already embedded at the base of the actuator to actuate the end effector. And I've got an example of that. ⁓ I think I had sent Leo the video. It's of this robotic dog that I thrown together. ⁓ where one axis is used to actuate the shoulder and the other one is used to actuate the elbow. ⁓ and all of its controlled just with one unit. So it's got this four axis leg just with one actuator, which I think is really cool. And then I know I mentioned this earlier. ⁓ there's, you know, generally thermal management problems with the way these three axis, four axis joints are usually designed in humanoids. It's really obvious if you've ever watched the races that China hosts, I think the winner of this year's race. was the w was the guys that figured out how to do the thermal management in their robots, but a majority of them hadn't figured it out yet. And I say they figured it out, it doesn't mean that it doesn't look crazy. You know, you've got all these pipes and stuff sticking out of every single joint of this robot. But because with the the orbit actuator, all the motors are down here, if you're housing it or you've got it in your your your robot, well all you gotta cool are the motors back here, right? You don't have to worry about cooling any of the motion that's happening at the joint. So you don't have to worry about all these moving Cooling components. And then, of course, outside of humanoids, it has lots and lots of different applications. I'm viewing it as like a third primitive. I can be used as a platform technology to develop all other kinds of systems. Now, in terms of downsides, and these are things that I'm currently figuring out. So I don't think they're permanent downsides. One, I'm working on the DFM of the design. Right now it's plastic. Now I have good news. I recently started working with two really incredible people. ⁓ one of them is Peter Middleton. ⁓ he runs a company called Bishop Aerospace. They do a lot of ⁓ stuff with like military. He's gonna be helping with the DFM and like manufacturing the first metal prototype and potentially demo units down the line. ⁓ in addition to that, I'm working with Tyler Chadester. He runs ⁓ Metamorpheus Technologies, but prior to that, he was the former head of motor design or motor manufacturing at One X. So he's gonna be working on the motors that are going inside the actuator. ⁓ and he's also helping with the rest of the internal design to help fix all the little tiny issues inside of the actuator. And one of the questions that I get in terms of downsides is like, how does a singularity work in the actuator? And you know, that's something that's present in pretty much every single ⁓ multi-axis system. And how we're gonna get around it is mostly just through software, right? So either it's like a path optimization thing or you know, you just avoid the very, very pop point of the joint. You won't have to deal with ⁓ any of the issues that you normally see like. singularities and multi access systems. Now you guys tell me if you see any downsides. ⁓ that are like plainly obvious that I'm completely not mentioning. I would love to hear what you guys think.

speaker-0: I don't see like really like downsides, but just what a question that just came up to me when you're talking about this collaboration with and these other guys. Do you have like a team that with with some co founders or are you like on your own now or is it more like an open collaboration? Or what is like the setup right now?

speaker-2: So Tyler and Peter are both getting a little bit of equity for their help. I did have co founders at one point. One of them is named S ⁓ Nick Kill. He was the P in C and P R and D. The name is tentative. I get to work on changing it eventually. He wanted to go focus on his career and no bad blood or anything between us. I totally respect the decision. I get it. ⁓ startups are hard. And ⁓ he was trying to balance his day job with this and it was just surely a nightmare. And then we had another guy that was in sales, but we just decided, you know, it's too early. So there's no real point in keeping him. ⁓ so right now it's just me. And you could count Tyler and you could count Peter. They they've been great so far. Yeah.

speaker-0: A question back to the downsides. I mean, ⁓ I'm just like wondering, okay, can you scale this infinitely or is there like a limit? I mean, you said already materials. I guess there will there are some implications if you use something else than what you're currently using.

speaker-2: Surprisingly enough, okay. So the the scaling the actuator is not that hard. It's the only real limitation is the size of the motors and the size of the bearings we're using. Well, assuming you can accommodate those, then you can make it. As small as the knuckles in your finger if you really wanted to. Or ⁓ as big as the the joints that you'll see in the giant Yaskawa robot arms ⁓ putting together cars. And I think it would just it would be just as effective across those different sizes. Additionally This is a different way to look at the scale of it. Because of the way it's designed, it's very easy to add additional degrees of freedom to the joint. So you could theoretically keep adding additional motors on the bottom, and then it would correspond with additional axes of motion up here. So you could have a six, seven-axis robot arm wherein all of the motors are mounted at the base and the arm itself weighs almost nothing. ⁓ but it has the torque of what you would normally see in like a a big hefty robotic arm. And it wouldn't have any of the downsides. I'm sure it would have its own, you know, demerits in some capacity if you have this many motors stacked like this. But ⁓ you wouldn't have to deal with the the typical downsides of like a multi-axis arm.

speaker-1: Yeah, for sure. I I think the base mounted mounted motors design is like the smartest thing about it to me. But I I'm curious, you don't seem to communicate on talk on your specs. Yeah. Is it intentional or

speaker-2: So ⁓ that is really just a consequence of the motors we're using. ⁓ for the sake of prototyping, we're just using three off-the-shelf motors we bought from China. It's a big holo shaft motor that we bought from Steady with. ⁓ each of those, I think they have a peak torque of ten newton meters a pop. ⁓ right now it's all direct drive. There's no real downside in terms of loss and torque of the gears are. While they may not be the most manufacturable at the moment, they're not really that inefficient. So you don't have to worry about the torque dipping or anything. But it is pretty easy to integrate reducers into the design. It's just a matter of sticking reducers on on top of each of these motors. And the reduction is going to transfer over to the the motion up top. So right now, each axis is ten newton meters of torque, but Tyler's cooking something up. That's going to be really crazy. So I'm looking forward to getting a spec sheet out there to everyone that wants one and there are a lot of people that want one. I promise.

speaker-1: Very cool. Yeah, I will be very, very eager to see that. So yeah, I looked a little bit into ⁓ the history of spherical actuators and there are precedents actually. ⁓ you probably know about them. So there is a professor at Georgia Tech, so he's pretty close to you. ⁓ he's called Cork Mengli, and he has done research on permanent magnet spherical motors. And I've also seen that ⁓ Carnegie Mellon. Has a history of doing spherical induction models. So what I understand is that ⁓ these models, they are, you know, they use magnets, you know, to activate, while yours used geared ball joints. It's important to underline because I mean we are going to display that for the people who watch us, but for people who listen to us, ⁓ you actually have gears inside your high graders, and that's what makes it move, and that's what makes it powerful, right?

speaker-2: Correct, correct. Yeah, we've got ⁓ a series of bevel gears and a ring gears, ⁓ that allow the motion to happen.

speaker-1: I see. And so my question with that is you have three models in you know in the base. Does it make it as many points of failure? What does serviceability look like? Is it better, worse, the same than having the three stacked act traders that you mentioned that already exist in traditional ⁓ humanoid design or the tutorial? So

speaker-2: You saying like is it as good as what's on the market, basically?

speaker-1: ⁓ not necessarily as good not I would say as reliable and when it comes to serviceability, for example, if one motor breaks, does it break the entire actual or only one degree of motion? Things like that.

speaker-2: The actual is incredibly easy to service, and I would say more easy to service than the typical multi-axis joints you see. In order to say, swap a motor out, ⁓ all you really need to do is is take this cap off right here and then take these ⁓ bolts off that go around the diameter of this chassis. You can take the motor off, swap it for a new one, and that's all. ⁓ every single one of these motors rotates independently of one another. In relation to the three axes of motion of this joint up top. So you don't need to worry about a single motor destroying the entire actuator. Now, if for whatever reason all three motors decide to ⁓ explode, you can very easily swap out the entire unit for another one because the design is incredibly modular. And then I think the other part of Leo's question, how many points of failure does it have? Now I don't have an exact answer to this. ⁓ but I can tell you that it is less than what you would see in a typical serial axis joint. In terms of the I know you were referencing the permanent magnet spherical motors. Those ones are really cool. They have quite a few issues though. One, it's it's really difficult to do anything that it's like bearing adjacent with those. So it kind of needs to float right in the center of this spherical array of magnets. Are it electromagnetic coils, which as you can imagine, causes some problems, especially in terms of torque. And also, because of this super complex array of electromagnets, you run into a lot of distortions with the magnetic field that you're using to control the system. So I'm not confident in the precision of those, especially under load. And lastly, you really run into a limitation in terms of the the work envelope of these ⁓ permanent magnets. spherical motors because it it is a ball inside of a socket and the socket is these electromagnets, you're limited to the size of the socket, right? Especially if you want to entirely encapsulate this ball to get the best motion you can get out of it. So it's it's kind of it's conical, right? The work envelope is conical. So it's not super practical. And also they're heavy as hell. I mean, I don't know if you've seen them, ⁓ you're gonna put ⁓ on the screen, I'm assuming. ⁓ they are d enormous. And I don't know how practical they would ever be in any actual mechanical system. Now I don't doubt in the future there's gonna be some genius that's able to figure it out, but you know, I haven't seen anything yet.

speaker-1: Yeah, I'm guessing that's why they remain a research topic and they never really made their way into the industry. And this leads me to another question, which is would you do the entire production in the US or would you have to source some parts from China or from Japan? Because you know, we talked about the recent trade ban on Chinese humanoids in the US. Also recently, Standard Bots raised two hundred million and standard bots wants to manufacture their entire robot arms end to end in the US. So on which side of the debate would you

speaker-2: Beth. Ooh, ooh, that's a good question. You know, I'm not sure yet. Because the design has yet to be or because we don't have a product completely developed yet, right? ⁓ it's still in the prototype phase. The final components that we need in order to design the product have yet to be decided. So I could totally see it being either which way. Now, does that mean we'll go with China? I don't know. I know I've been talking with a few other companies, like Shaffler is a big one. But I've been talking with they're based in Germany, right? ⁓ so we might end up sourcing components from them. So I don't know. There's there's a good chance for either one. Now I see the benefits of both. I also see the negatives of both. Working in America, obviously you you want to foster the industry that's kind of been demolished by China over, I mean, more years than I've been alive, right? But also China's manufacturing capacity and their Complete breadth of components is crazy. And the price of each of those, I mean, you can't beat it. So if we're arguing if if we want to scale fast, then you would go with China. But there's also the risk that if you work with China, then the FCC might throw something in your face and completely curtail your entire business model and your entire production line. And then you'd be screwed. So I'll put an asterisk and say, we'll figure it out. We'll see.

speaker-1: That's a good mindset. Because yeah, it becomes a structural risk for most robotics companies right now. I mean, for most industries in general. I mean, my last questions are around your market and your customers. To give you an idea, agTraders are usually approximately half the price tag of each humanoid, right? Because they are expensive to make and the humanoids had lots of them. So I'm curious about who is your your customers? Do you want to sell your ag traders to robot manufacturers, would you license the architecture or would you develop the product and hope to be acquired? What is your mindset of that? If you have one, maybe you are very open in your current phase. So

speaker-2: I don't know if there's one particular route that we're completely dedicated to. I think it's gonna be a a mishmash of ⁓ several. Early on, I think the best route to go is the licensing round. Only so we can bring in enough capital to sustain or to create a, you know, manufacturing or create the manufacturing capacity we would need in order to sell these at scale. So once we do license and once we bring in some capital, then we will do the manufacturing in-house and then sell to the ⁓ these OEMs or these these these component manufacturers to be used in different kinds of systems like the robotic arms or the humanoids or the robot dogs or whatever. So it's it's really those two at the moment. We'll kind of see I've I've got a bunch of stuff cooking in the background, but nothing's been completely fleshed out yet. ⁓ there's some potential partnerships that may end up working out and then that would kind of decide the course of Where we would go. So we'll see.

speaker-0: Yeah. Can you name any partnerships you have that you're working on or is it like confidential?

speaker-2: Cool. ⁓ that is a good question. Tell you what, I'll I'll tell you and then I'll let you know after the recording if ⁓ you can keep it in or not. There's one of them. So we'll see how that goes. I was working with my actuator, the company, not it's their name is weird. I was working with them for a while, ⁓ but that didn't end up panning out because they got bogged down by ⁓ creating a new manufacturing facility, so they were just hellishly busy. ⁓ so that's something that might be picked up Again in the future. I don't think I can say the other one, so I'm gonna be on this.

speaker-0: Probably also talk to VCs, I guess, trying to raise capital. How how did that go? Can you talk about that? Ooh.

speaker-2: Okay, this is a fun one. So I have been stuck in a chicken and egg situation for quite a while. So and this has recently, I think, been alleviated because of Tyler and Peter, and thank God they're helping now. But prior, all the VCs I discussed or talked to were of the same mind. They wanted me to come back once I had a product, right? You you have traction or you have a product that we can see fully fleshed out. We don't want to see a 3D printed plastic thing. Like it's a really, really cool concept. We love the idea, but come back later. ⁓ and then I'd be like, okay, so I need to develop this product. ⁓ what do I need to develop a product? Money. so you can kind of see the issue here is in order to get VCs' attention, I need something to show them that other people can buy. But in order to make the thing that people can buy to show the VCs I need money from the VCs to you see where the problem comes? Yeah. ⁓ now that isn't to say that the the VCs were all interested in the tech because I think they can all see the vision. Way I I look at it and I think I'd already discussed this is, you know, this is this is a a new primitive. ⁓ there's a reason that so many people have tried to make something like it. And I I genuinely think that it could be something on the scale of rotary and linear actuators. And that that idea, just the sheer magnitude of the potential, really caught the eyes of these VCs. But all of them are incredibly risk averse, especially in hard tech. So, you know, that's where that's at. We'll see.

speaker-0: But how much how much money is actually missing? Because I mean, do they expect a full fledged production line or just like a single unit? I mean, how much money do you need? Maybe maybe there's someone in the audience, ⁓ who who can support you?

speaker-2: Right, right, right. So I I think for just w like a demo unit or sending out demo units, I think something in the the ballpark of like two hundred fifty would be good enough. If we wanted to completely flesh out the design, hire a team, maybe do like a small scale manufacturing. I think that's in the ballpark of like two to four million. Just because hard tech is so expensive. You need that kind of capital, right? It it's not like a software company where you can just ⁓ have Claude, you know Five code you a a website and an application. You know, you gotta pay for all of the raw materials, you gotta pay for the motors, a lot of times you gotta pay for the tariffs, all the electronics, ⁓ machining costs, labor costs, all that jazz. So ⁓ two fifty for demos, two to four million for, you know, the whole package, we'll say.

speaker-0: think a interesting thing that you mentioned was that you were considering the licensing. Because I re if I remember correctly, the founder of Neuro Robotics, you know, the German robotics company. I think he has a very interesting ⁓ past. Which I also go to in a episode of us. He had the a similar problem when he wanted to start in the robotics company. He had basically nothing and ⁓ he started with licensing as well, as far as I know. Pretty interesting. So I think there are many ways and many routes that you can go with this and especially in the core technology is fledged out, which it sounds like I mean that's ⁓ it's a pretty remarkable tech and then looks ⁓ it looks kinda satisfying as well when you move it around to be honest. Yeah. It's always got it got like a little fidget ⁓ thing going, right?

speaker-2: It's very fun. It's very fun. I'm sure. I mean, I sent Leo a bunch of videos. You can probably put those on screens to like an ASMR thing in the background or something. gotta

speaker-0: Fourteen hour ⁓ action.

speaker-2: Yeah yeah.

speaker-0: Yeah, that's cool. Yeah. Maybe going more to the general outlook on the robotics market, what are your predictions? ⁓ I don't know how much you look at models and ⁓ capabilities in general, how much you look you look to the left and to the right because you're probably laser focused on your product. But what do you where where do you see robotics in the next five and next ten years?

speaker-2: Ooh. Okay. ⁓ so I'm not gonna go the Elon route and say that everybody's gonna have one in their house. ⁓ I think that's ridiculous. No shade to Elon, but I yeah, whatever. I think it will start dominating a lot of the industries that rely on really menial, almost depressing labor. And I see the potential issues with that. Obviously, people are not gonna wanna lose their jobs, but I also see the benefit. A lot of the dangers that are associated with those jobs and, you know, the potential deaths ⁓ that come with those will no longer happen. ⁓ you can weigh that how you will. Now, in terms of it being in the home, ⁓ and I'm all talking about humanoids, not just robotics in general. I could see it working if they really get down the ability for it to do chores well and consistently. Because fundamentally nobody wants to do those and they're buying the robot to do the chores, right? They're not buying it to take care of the children. They're they're buying it to do the dishes and the laundry and back in the floors and Whatever. And I I don't know how soon that will be, but I definitely know that nobody's gonna really adopt it if you're still relying on teleoperation for these robots in the home. I don't know if it's just me, but that that seems creepy. I definitely would not want that in my house. ⁓ Yeah, yeah, okay, okay. I do definitely see it being adopted in a lot of industries. ⁓ the potential's there. I mean, this is an argument that a lot of people say.

speaker-0: We had those discussions, yeah.

speaker-2: But anywhere where humans do labor, you could literally just drag and drop these humanoids to do the same thing, right? What the time scale of that looks like, things accelerate fast. I could see it happening in maybe maybe the next seven or eight years. But I definitely think we still have long way to go from all the demos that we see nowadays, which is robots punching each other and robots dancing and doing the Dougie to, you know, what we need to be seeing. And to be fair, there are some companies that have really developed, you know, actual capacity to do meaningful tasks. Figure did a pretty cool demo with this robot on the assembly line. ⁓ I enjoyed that several day ⁓ escapade watching that on on YouTube. I think Boston Dynamics is doing a cool job. I mean, that could be me fanboying, because I mean, who doesn't over at Boston Dynamics? ⁓ but their work in the the Hyundai facility is really cool. Hopefully that answers that question. I went on a bit of a a tangent there.

speaker-0: I think everybody's is feeling that, especially the ones who have like some access to this ⁓ sector. Maybe another one more in the ⁓ theoretical sphere. What do you wish people worked more on in the robotics space? What is something maybe that you have to encounter where you're like, come on, why is nobody doing this? Or maybe something that you would have considered working yourself on and you're like, nah, I'll prefer working on the X Rator. That is

speaker-2: That's great question. I mean, the cop-out answer is like people should be making better actuators. ⁓ but I'm not gonna say that. I I think exoskeletons have an incredible incredible amount of potential, especially robotic exoskeletons, to help people that have like disabilities or are like working through health problems or anything like that, both in an assistive and a resistive capacity. I mean You could have these exoskeletons help you walk, help you walk again, you know, give you back that autonomy that you lost. And this next part is gonna be based on the whole selling point of the one startup I did a while ago. You could also have exoskeletons that are resistive to your body and kind of allow you to exercise wherever you want to be. ⁓ just doing random menial tasks at your house. You could wear this thing and that would resist your motion and basically Work yourself out. ⁓ it'd be like a a bow flex. I think that's the name of the machine. Just on your body. I think that'd be dope.

speaker-0: There are these ⁓ extra skeletons, right? For for walking. I think there's one that you ⁓ get around your hips. And I think there are two ways that you can put it and and one is actually resistance. And I'm not sure if people actually use it, and it it still looks a bit clunky sometimes when you see it on video. It's still it's like it's not like entirely smooth, but ⁓ I I can see that as well for training, for example, or things like that. ⁓ I saw a video of a guy turning it up and cranking up the resistance and he was walking like Like an old man basically. So ⁓ but yeah, maybe you get some muscle.

speaker-2: Yeah. And and I guess my other answer, and this is kind of tangential to what I said earlier. I think prosthetics could really use some attention from the the robotics space. I think it hasn't had a lot. One, because there's kind of a big monopoly that kind of dominates a company called Autoback. ⁓ they kind of dominate the industry and they really prevent a lot of the innovation that you would otherwise see. Also, the industry is relatively small. So a lot of the companies ⁓ are, you know, that they've got. investors and those investors want them to go after things that bring in a lot of money. So these people that would otherwise see the benefits of this incredible innovation don't get to see it because they're like a small group of people. ⁓ so I definitely think the technology that's being developed in humanoids could just as well be used in ⁓ the prosthetic

speaker-0: It's a very encouraging industry, to be honest. ⁓ there's so much potential there and ⁓ yeah. Maybe Liu, do you have any other questions?

speaker-1: No, ⁓ I asked all the questions I wanted to take.

speaker-0: No, I think Bryce, that was really nice, very insightful. I mean, we got it all, many different topics, but of course also getting first hand experience with actuators was something I think we have to have on this channel, and ⁓ it's gonna be a staple on this channel. First interview as well. ⁓ that was a tire. Thank you so much for watching the robotic stack podcast, episode eleven. That was Bryce Genchoto, Oscar, and bye-bye.

speaker-2: Yeah, I'm honored. Thank you for inviting me.

speaker-1: Leo.