HAS 1X TECH REVEALED THE 'ULTIMATE' HUMANOID HAND?

[00:00:00]So for 70 years, robotics worked around the hand problem. The humanoid bet is the reverse. It lives or dies at the fingertips. That's Burnt Borage.

[00:00:10]Speaking, not me, the founder and CEO of 1x tech. They've just unveiled the neohanoid bought hand. 25 degrees of freedom tender-driven hands nearing or surpassing human level dexterity, strength, speed, and reliability. This was what they teased us with yesterday and it set the cat amongst the pigeons.

[00:00:32]Alienware is what we called it. Um, and we had our very own Scott Walter respond to it. And as usual, he was right. I'm going to bring that up. Scott, I wanted to just give you a shout out straight up. Right out of bat, you were right. As always, are you going to gloat now insufferably?

[00:00:55]All right. So the rock paper scissors gang is Spock gang sorry Marad don't want to leave you out is back and we've got a special guest Devong Adaru from the humanoid hub is with us today.

[00:01:05]Welcome Dang. Uh Devong welcome gentlemen. Great to have you all with me. We've got um quite a task on our hands. Alienware is what we are going to break down from 1x tech. So come on, Scott.

[00:01:19]>> I I think kind of start out is that this kind of at least in my time time zone came out like yesterday evening. So that's what I spent all my evening doing and then then today I had other things when the it went live. So, we are basically going to see how the sausage is made on this podcast right now because um I haven't really had time to give it full serious thought beyond what I saw here last night >> and beyond what I mean what will probably be happening is that if you see me reaching for my phone, it's probably because I'm getting a text from one hand expert here or there that wants to weigh in. Um, and I definitely wanted Devong to come on because I have a feeling of all of us, he may have spent the most time actually looking at the the full representation, but to begin with the teaser, I mean, the first thing we always do again is like count how many tendons there are.

[00:02:07]>> Yeah.

[00:02:07]>> And then from that try to imply how many there may be. So, we can see I think 20 in this view and then we're assuming there may be another 20 in the back because it looks like there could be a nice rotational symmetry around here. We could be right or wrong about that. So there's either 20 or 40 >> ring a better view. Okay. So this is a better view. This was shared by >> Okay. Now, now it's kind of like giving away a little bit of the answer, but I haven't had a time to go through and like kind of verify that.

[00:02:33]>> So when I was thinking of like, you know, is it 20 or is it 40? It does the number 40 tendons uh sound familiar?

[00:02:41]Does does that ring a bell?

[00:02:45]>> Number 40 tendons.

[00:02:47]>> Yeah, Kyber had 40 tendons. So, so as soon as I'm thinking 40 tendons, like, oh, if they have that many, they might be using a similar actuation strategy.

[00:02:55]Um, the the other thing would then be the number of actuators, and we can kind of see what appears to be rings of five going around, and it looks like there's at least four going down. So, that would have been kind of 20. And clearly, we could see that the tendons were being paired at each actuator. So, that means that when when one was like kind of in tension, the other was in compression, and that you don't need to have a dedicated actuator for each tendon. So if you had 40 tendons, you don't want to you don't need to have 40 actuators. So I was wondering if it was like a similar strategy, but as we've seen a little bit more, some things have sort of changed um looking at it overall. Now the first thing that was obvious to everyone is the use of Bowden tubes. And if you don't know what that is, basically it's the sheathing around your bicycle cable, right? It's the exact same idea, right?

[00:03:42]And we've seen this before on the original um 22 uh do hand from Tesla >> where they were also using the Bowden tubes and we could see them kind of moving around. They then went away from it and I think just about everyone I know of that's attempted to use Bowden tubes has given up on them and at this point 1X may may be the last one. Now there are a few things that they seem to have done differently in the case of of using these. Now it may be Bowden or Bowden. I think uh everyone in the industry calls him Bowen. You know, it's like Bowden Bowen, tomato tomato.

[00:04:19]>> I'm just going to quickly bring up your posten.

[00:04:22]>> Yes. Yes.

[00:04:24]There we go. Okay.

[00:04:26]>> And so the the first thing was that that would >> I was a bit rowdy when I was younger.

[00:04:31]Yeah.

[00:04:31]>> Right. Now, if you look really closely in in the tube, so they are a little bit transparent. Uh for those you had a bicycle cable and ever had to take it apart, you'll notice there's like a kind of this plastic out part outside and in the inside there is a metallic part which is basically looks like a big spring that gets all twisted up and if something happens to that thing and like it goes all is like it's game over. You got to buy a new cable, right? And normally the easiest way to make a spring is is with a round wire. You just twist it around there and you don't have to worry about lining it up. Um, but what happens in in compression is that as that spring comes together and those round parts are on there, things kind of slip and it doesn't work really well in compression. And these tubes are under a lot of compression. So when you're putting that tension um on there, those things are being compressed down trying to do the routing and everything. And one way to be able to accept the compression a little bit more and actually give overall better rigidity is to actually give those wires a square cross-section and twist them around that way. And and that's one of the things that seem to be noticed right there is it seems to be a square profile. Now have actually Mirdad does that square profile sounds familiar when you connect it with one X or basically >> previous name.

[00:05:46]>> It it's funny with the maybe um previous name Hollow or >> Hodi. Yes.

[00:05:52]>> Yeah. But but also uh recently I've seen three hands with square profiles. I don't know. It's becoming more trendy apparently and I I want you to explain why. Um there are some reasons but it seems to become more popular at some point.

[00:06:07]>> Yeah. Yeah. Yeah. Well, I mean when when it comes to the ho motors, it's because uh with the windings, you want to com have as much copper in as tight an area as possible. And circular profiles do not pack as well as square cross-sections. So you can see that at the supermarket that usually if you want to have a lot of milk in the same place in the fridge, you want to have square containers and not round containers. Um, and so that's so if you want to pack as much in there, especially when you're making motor windings, that's very important. But there's obviously certain tech that you need to know and how do you take something that's a square profile and do all these twisting and bending and make sure they line up right. And that's why most people give up and they say, "Ah, heck with it. Just keep the whole thing symmetric." And then as things go around, it's a lot easier. So, if you're making these windings going around, it would just be a little bit easier. Things will maybe uh rub or move over each other a little bit easier. But the thing is that that the problem is compression, right?

[00:07:01]Because two circles are coming on top of each other and it's you're you're contacting at a point. And if you have a square cross-section, then you got a flat on a flat and then compression is a lot better. Uh which means you don't get any deflection. So, it's you're you're when you go into compression, it's literally not going to compress. it's just going to kind like stay rigid and not move. So, it may have been that they said, "Hey, we we're comfortable figuring this thing out and we'll do that." And I don't think you can source those tubes like that. My suspicion is that it may be something they actually um built internally >> uh to go ahead and get >> Oh, interesting. Okay.

[00:07:35]>> Yes. Now, now >> I thought they were off the shelf. No.

[00:07:38]>> Uh maybe, maybe not. Okay. Uh, there was one comment underneath my post about that that said they were round off the shelf >> by someone who may or may not still be a 1X employee um who who had some some knowledge that that he he was able to make that claim only to then be um for to be contradicted by uh by Barren right after that >> and that actually Scott is correct.

[00:08:07]So, and I think that that, you know, inferring to that that that that's actually what they're doing with the Bowden tubes, Bowden, sorry. Uh, and that's one interesting aspect that, let's say, maybe is the the key or the unlock or the trick to going back to using them because everyone abandoned them. You know, I've almost abandoned them on my bicycle because I I know that after a while they wear out and sometimes they literally explode on you uh if you put like, you know, if you break too hard or you do something like that, there's there's too much tension on them. So, that's a tricky part and that allows them to do this routing back and forth.

[00:08:41]Um, and then I think the other thing you can see is uh you can clearly see the uh what the tendon material is. So, if you if you stop the movement, you can actually see the threads being tied off and you can see how they're clamped down. So yeah, so when you go over to the other image, so that was looking at the Bowen tubes and there you can sort of see it coming down. You can see how they're clamped in. They're clamped in place there and um you know that definitely looks like some kind of polymer thread probably daema u exposed.

[00:09:13]>> Yeah. Yeah, that's that that's exposed.

[00:09:15]Um u at that point it's okay to expose it. It's when you come out the top part that it's going through the tube and that is something that you would be able to tug on if you need to retension. So there it doesn't look like there's auto retensioning but looks like it's something that would be done manually.

[00:09:30]>> And um it's a question to blend.

[00:09:35]>> Oh, dyna is pretty um pretty flexible.

[00:09:39]>> Okay. Um, and the question is is the size of it is like obviously the thicker you make the thread, uh, the more careful you have to be of the radi you kind of go around. And that's part of the idea of the Bowen tubes is it kind of routes everything very gently around there so you don't have any really tight uh, radi. There are a few places, as you can see, especially when they come down to the bottom, it's making a lot of these S kind of shapes to be able to route themselves down to where the lower base uh, ring of actuators are. So that's like the but still they're very gentle curves that are going along through there and and Dynam is usually pretty good on it. But anything that's a a small radius is going to result in a lot of wear either on that radius or actually on the dynam itself. And in many cases everyone is trying to use like a um either a very gentle radius or some kind of pulley system that they actually have a a rolling contact as opposed to a rubbing contact on there.

[00:10:37]Um, and the other thing is that you've got this trade-off between if you want this to last really long, you want to have probably the the thickest thread possible.

[00:10:44]>> So, you know, it might be in the order of a millimeter. The problem is the more that you have that, then you have to be more careful about the radi that you're going through there. So, is that that little balancing act along there? Um, but >> what it can do is the thicker it is, the less likelihood you're going to have of creep where creep is sort of the sign of failure that you're going to have and the maintenance you want to have. And when I spoken with with Barrett before you I think they talked about this having something like two million um cycles or something like that that that it's able to go through sort of meantime between failure >> and that's achievable so long as you stay within what are the uh the nominal limits of of what it's rated for >> and in many cases you can go above that and sometimes people will say >> you know the nominal might be uh you know 10 newtons but rated for 30 or something like that and then everyone says oh that means I can run it at 30 and you run it at 30 and you have like one or two orders of magnitude difference in in the lifetime of something. So the idea is that to uh to kind of operate well within like the middle part and then you will get really long lifetime out of it. But if you're occasionally exceeding that and doing it a lot then you you know you're going to see degragation almost like that. So those dynamo threads um aren't they going to interact with the the sheathing material whatever is used um >> yes which means usually you need to make sure there's low friction in there and the other thing is that is that sheathing that's going around there has very gentle arcs and curves going on there so you don't have anything sharp you know the the one concern I have with a square cross-section is that you've got uh edges in there or or or sorry corners and nature aborts a corner as we all know and that whenever you have something that's a potential source of friction. So if they're compact and they're really nice they may not be exposed but if you just get like a little bit of movement out of that you could expose something and that's something that one of the threads could rub up against. So hopefully that's not happening. That's I think why everyone goes with more of the circular cross-section because you're always rubbing up against something that's smooth and doesn't have a sharp not quite nice edge but you know something that's good enough. So it's like engineering is a tradeoff.

[00:12:56]>> Okay, I quickly it could be that they put like a lubricant in there which you know can kind of protect against that.

[00:13:04]>> So uh quickly before we get to the um bio biomechanics and biomeicry with Gustav Dewang your uh quick highle view of uh what this is um you've had a bit of time to go through it. What do you make of it? Yeah, I was just uh I have I was just amazed by uh the willingness to just show it to the world and there they've shown a lot of guts in this one.

[00:13:32]>> Yeah, you're right.

[00:13:32]>> And and uh >> I I think Oliver Stark's reaction was they really opened up the kimono and that's unusual.

[00:13:39]>> Yes. And yeah, before this I would have assumed they were using linear actuators whether I don't know whether Scott has opined on it or not but this one seems suspicion was was always was probably going to be rotary. I think that's what we'd sort of seen before. I mean, you would think, >> but in order to compact them in there, the rotary's >> Everyone's having a hard time with the linear, so I'm not surprised they did this.

[00:14:03]>> Um, and and you have the other advantage with the rotary is is that you can tie it off on both sides. So, you get that capsson drive, which is something that um there 1X is doing everywhere else in Neo. So, I'm not surprised that they would apply the same. So the capsson drive is with the one that it's like acts like a pulley.

[00:14:22]>> Yes.

[00:14:25]>> Exactly. Exactly. That way you get a two for one and it's you can kind of do it with a linear drive but it's really tricky and that your linear drive has to stick out at the bottom part then so you can have something around there. It's kind of pulling on one side.

[00:14:38]>> What about slippage in that pulley structure?

[00:14:42]Oh, it's tied off so you shouldn't that the main problem you have here is creep >> and that and that that is it extending over time >> something that like you have in the planet free rollers um but it's so tightly packed in planet free rollers that not not planary rollers what what is the other one that you had reviewed in uh in Europe um >> well they're planetary roller screws that that would be the linear drives or are you thinking for rotary or linear drives >> um rotary >> rotary So there'd be planetary and then there would be um harmonic drives. Um there could also be cyclloidal but these are most likely like planetaries.

[00:15:20]>> So they they could be one or two stage planetary uh in there. Um maybe like around 15 millimeter diameter something like that is like the kind of the guesses that I've been seeing and maybe 1 millimeter thread somewhere around there.

[00:15:38]>> Okay.

[00:15:40]And so now I think the biggest concern everyone has is that um it's really hard to model what's going on here because when when those bowen tubes are going all over the place, it's really hard to tell um what the signal is that's coming from. That's why everyone would prefer to have like a straight shot with your tendons >> to keep them kind of under a proper tension the whole time. It's a lot easier to control. So mastering this is a lot tougher >> uh than than it would be because the the it's adding a a lot of kind of additional friction as well as strange tensions and tuggings to all the threads going through there.

[00:16:22]I mean they're using a lot of similar language as Kyber is with the force transparency and that they can kind of know exactly which joint is due to the >> due to their actuators I guess. So maybe they're on to the same thing but what you're meaning is that the the tubes are kind of moving left and right.

[00:16:42]>> Yes. Yes.

[00:16:44]>> And and and the only thing that complicates that is that u is the wrist.

[00:16:48]So they have at least two degrees of freedom up there.

[00:16:52]>> Uh and just the movement of the wrist itself is going to cause the Bowen tubes to deflect >> and and just that deflection itself is going to send some kind of signal which would be kind of like a fault signal >> and and potentially cause stretching as we know happens with our hands. my I I haven't I've been trying to look at it close enough because the language that they say it's like 25 >> doss >> with basically three in the rest and I don't know if they mean three at the carpal joint or if they mean two up here and then one down at the elbow >> which I I would suspect one down at the elbow because it just makes your life a lot easier.

[00:17:30]>> Uh but the mechanism that they have up there looks a little bit interesting that we can't see here that we can see in the full one. Um, I suspect that that's probably too. So, we're seeing the yawing there.

[00:17:42]>> And then are we seeing any We're not seeing pitching yet. I haven't seen the roll yet, so I don't know where the roll is actually being done.

[00:17:50]>> So, it's kind of this design is very similar to the deck hand by Rob Knight because it looks like what that design had was four metacarpals that were like fused along a pin. So, it could flex and extend along that pin. Here you can only see two which I think is like the second or combined with the thumb and the fifth. There's like a gap in the middle where the where the >> Yeah, let me show you this one instead.

[00:18:14]>> That's where the the Yeah, there you can see it closer. And the similar also what they seem to either they're thinking in the same way but a similar design is that the the fitting for the thumb is towards the third metacarpal. So the third metacarpal then thumb that will give you the thumb in a nice position because our adductors are coming from the third metacarpal and not from the second. So it's a good positioning of it. I would hope that they put the thumb the basal joint of the thumb a little bit saddle joint a bit further up. It's a bit close to the mid part of the but it puts it in a really nice position to get into opposition I think. Um, but a lot of the design here, first of all, the movement. It's kind of like they're playing around with the the Tesla bot they found in the bushes. It's like doing the same motions, but it's not as in in what you call it inaccurate when it comes to the motion. So, as you were saying, when it flexes and extends the the wrist, the fingers are not affected by that in the same way that the Tesla hand was in the bushes. When then we then they did the kind of ab and adduction of the wrist, the fingers would kind of flex. And I don't see that in this hand. So they seem to have better control of the fingers as the wrist is moving, which is impressive.

[00:19:28]>> You got to stop saying Tesla hand in the bushes because people are going to misunderstand you.

[00:19:34]>> The old clip. Not everyone is there as you. But you can see the fingers are perfectly straight when it's doing the >> Yeah. Yeah. Exactly. So what's going around is not the it's not inducing any sort of movement in there. And and I think that's something that the Bowen tubes should be able to to do for you >> is is allows you to do that without having the the lengthening and the stretching if you used a comb or some other kind of mechanism there. So many of the other hands have sort of given up on the yaw because that's what really causes the problem. So you can compensate for the pitch fine. You can set up something there but as soon as you have those two degrees of freedom then everything kind of falls apart and and that's why Bowen tubes is kind of the solution for that. Um, now in the counting, are you seeing three degrees of freedom at the CMC joint?

[00:20:22]>> No.

[00:20:24]>> It's really >> Are you seeing five total for the thumb?

[00:20:28]>> It's rotating a lot. It's not really doing I haven't seen any of these clips.

[00:20:33]Maybe it's there, but I don't see the the flexion extension in this joint. I only see a rotation down here. There is no other movement than a rotation at the basal joint. And then you have these two doing flexion extension. But that might just be me not having seen it clearly enough. But this is primarily looks like it's rotating down there. And it might be a separate and adduction joint or pitch jaw flex extens or I don't know your lingo >> with more movement here. But I mainly see rotation down at the basil joint or the saddle joint. And then >> but if you look close if if you go to that baseball card >> uh that was showing the stats um I think we get a a more zoomed in view.

[00:21:15]>> And you know when I'm looking at that I'm I'm able to sort of see what appears to be maybe a couple of different joints there.

[00:21:22]>> Yeah.

[00:21:23]>> And and we have only uh the fifth metacarpal.

[00:21:27]>> So we we've got to get down we've got to figure out where's the 22 from. So we know they say 25 but three are the wrist. So that gets us down to really a 22 dove hand. We now have the fifth metacarpal which is you know usually the extra one. The the other 20 of course are the standard movements of the fingers.

[00:21:45]>> So each of these should have four and then the thumb usually has four. So you add all those up you get 20. And then we get the metacarpal that's 21. So it's like where's the spare one? That's why I'm thinking oh is it a five do thumb which some go >> for the rotation then? Yeah. Yeah. I'm wondering if they actually have like a rotation and if they have also the the abduction or flexion, whatever you want to call it, and then they might have the opposition.

[00:22:10]>> Yeah.

[00:22:11]>> And maybe So, that's the only way I can get to that count.

[00:22:15]>> You can kind of see the disc in that image that there's like a disc at the bottom of the thumb and that clearly rotates in some of the imact.

[00:22:24]Yeah. You actually have the mechanism just distal of that which would give your abduction adduction similar to the other fingers. And then >> I'm not sure but in if you zoom in on the thumb you could see that there are two tendons going to the first IP joint and then a single tendon going to the distal failance and it's the same for the bowler of the the palm side. So there are two tendons going up to here and then there's a single tendon going towards the next one. So there are two tendons and then the I guess it's the same.

[00:22:54]>> Okay. I I I was wondering yeah >> I sent the picture Roy maybe could pick that. I think it's a little bit clear about the tendons there.

[00:23:02]>> I I I think what you're seeing is is kind of the same actuation strategy that we've seen in Tesla and others. And Devong, you're definitely going to remember this when you went through the routing of that is that the uh the flexure is routed like right through the center of the MCP. So, it doesn't actually cause any flexion. So, it goes up and does the pip and the dip joint.

[00:23:21]And that's the one that you're seeing kind of exposed routing through there.

[00:23:24]And the other two that you're seeing coming up, I assume are just for the AB and the AD. And then behind it, we can't see the extensor. So my suspicion then is like four tendons per finger or digit. Uh and then so we would then go through that and then we've got, you know, 16 taken up right there. And then the question is like the thumb, how many we'd have there? Do we have another four to get us to 20? And then another two potentially that are then right there for the metacarpal and that might be the trick. Ah I think okay I think that's it. The the thumb the CMC here we go making the sausage. I think the the CMC joint is only two >> because remember it comes down to what are we defining as like degrees of freedom and everything. It might be that there's really 22 tendons here that we're trying to look at because I just said four four another four here right that gives us a 20 and then for the fifth metacarpal we need to be able to open and close it right oppose repose.

[00:24:33]So that means another two tendons are running out of there.

[00:24:36]>> So it's it's like this it's not really we're really we're not talking about joints. I think we were trying to count joints. What we should have been doing is really counting tendons.

[00:24:45]>> And then if you're if you're then going to tendons, there's 22 tendons.

[00:24:51]>> However, they're paired. That means there's only 11 actuators.

[00:24:56]>> So, >> take a look. See, are we only seeing 11 actuators there? That's possible because I was wondering how I was going to get another row in there.

[00:25:03]>> I don't think the actuators go all the way around the back now. Now, okay, there's a couple more actuators in there because we also have at least four actuators taken up for the wrist.

[00:25:14]>> So, we we we would need one this way, one that way, or at least tendons. We need So, we need just two more actuators in it, right? So, I think one this way, and then it will be able to reverse this way and that way, and that way. And I think they're using the same actuators for it. We I need to get a blowup of that and start tracing all them to see which actuators actually tie into which ones. Uh unfortunately Tesla really helped us out. Remember they color coded.

[00:25:40]>> Yeah.

[00:25:41]>> Yeah.

[00:25:41]>> To make assembly easier. And I think Barrett color your tendons. It'll make assembly easier >> for your Yeah. For your engineers. I mean it doesn't look like there are more than you can't really it doesn't seem like it would fit another bunch of those tendons going in the back when you look at >> and we don't see them I mean when we actually actually look up there through the um the carpal tunnel I don't see that many going through there I see them kind of thinning out and having a smaller number so if we look in there closely are we counting maybe about 18 or something like that going through there if four of them are already taken up so >> I mean there there are not that any >> you can see was kind of goes off to the sides each finger.

[00:26:26]I'm guessing the one going into the thumb to do the flexion is the same one that goes around to the back then. And then we kind >> Yeah. Yeah. Yeah. Yeah. It's So you can have two of them going in there. So the flexion and the extension are connected to the same actuator.

[00:26:39]>> So at the ball bearing on the thumb, you could see you can see a clear ball bearing there. And then there's like two strings. But if you go to the next the distal joint, there's only the one on the left in the image, so to say, right?

[00:26:50]>> So, I'm guessing the first one kind of just goes around that basel.

[00:26:55]>> Yeah. Yeah. Yeah. Yeah. So, so going on up there, you've got, you know, the ab ad. So, that's your two tendons going on up at the MCP, and then you got your flexion extension, which is another two that go up.

[00:27:06]>> And, and again, the routing is such that the flexion does not do uh the MCP. the flexion MCP is being done by pulling on both of the abductors at the same time, >> which >> is it's challenging. It's it's challenging to get that control.

[00:27:24]>> Um for the extension flexion as well if you have the same guy doing it, wouldn't it be tricky to kind of get it in the >> Yeah. Yeah. That's that's the thing is it's the trickiness and this is what what Tes the approach that Tesla took is that they're saying if we do differential, we get AB add A add, right? But if we pull them together, we get straight flexion.

[00:27:43]>> And then we have the second flexor which is going like straight through the MCP.

[00:27:47]So it doesn't create any torque there.

[00:27:49]And it gets routed around like this.

[00:27:53]>> Oh, you muted yourself.

[00:27:55]>> I accidentally muted myself. So So the pip and the dip uh are then coming from that flexure and then you have the extensor coming on back. the the the problem has been is that if you apply like that force here and you try to resist it with your abduction and that means you're basically pulling on this one and not on that one. So So now you have like a differential kind of going there fighting it but the result is you're also adding a little bit of flexion that you can't fight against >> which was a problem that Tesla had because they didn't have active extension. They were relying on some sort of spring return mechanism which was like too slow. So again, you couldn't do what you mentioned, Gustav, is like getting rid of the the mosquito or the fly.

[00:28:35]>> So, you know, you just waiting for the spring return. But now, if they have that because it's active, they have maybe enough antagonism that they actually can control that joint. And if they are, if they're doing that trick and they're doing it with active antagonism to sort of maintain rigidity there, that's impressive. I mean that that that means they they've solved some problem with the control >> because Kyber they have a separate one to extend >> they have a separate one right and and and what that means is there's absolutely no coupling so so they did it so that each IP joint is individually controlled and because of that they have no cross talk and they have really clear signals of where the um the force is coming from and they can do force control so they can when they they pinch down they can make sure the force is going off in the direction that they want. It's a little bit trickier with this. They don't have quite that fidelity, but with the active extensor, they may be able to, which is the basically has been the flaw in the Tesla design and what we believe they're changing in V3 because Elon's mentioning like, you know, the additional number of actuators they're going to put in.

[00:29:43]>> So, with this one, you're saying there is extensure which can which allows >> Yeah. at least one joint to go back and forth. Yeah.

[00:29:52]>> Yes. Yes. And and so we can kind of do that too. At least not we some people can some friends of ours.

[00:29:58]Shout out to Tom Zoo who who is able to do like weird things with his fingers because he's able to get kind of that that active control a little bit better.

[00:30:07]But for most of us we we have a hardware you can correct me if I'm wrong. The only reason we can do this is because the extensors are making sure we flatten it out. Otherwise, we want to do that.

[00:30:18]>> Yeah, exactly. So, we have or at least we use the muscles in here to kind of prevent the >> kind of do that.

[00:30:25]>> But funny you said about I think they show a small clip where the hand is really overexaggerating and it looks like like >> Yeah. Yeah. Yeah.

[00:30:33]>> First they they really made the opposite the retro position or retropulsion of the thumb which I've been talking about.

[00:30:40]They really show that but they also do it a bit over the top. I I can't say who, but but just someone else has retropulsion now and they're really proud of it and they want to make sure that you see it next week.

[00:30:50]>> Okay, good. But they also show the Tom's kind of hyperextension of the >> Yes.

[00:30:56]>> joint in this >> of the Well, yeah, the thumb was like, talk about like the hitchhiker's thumb.

[00:31:01]Man, that was crazy what they have.

[00:31:04]>> I don't know. Maybe it's Yeah, at least it shows a vast range of motion.

[00:31:10]>> So, the gloves.

[00:31:12]>> Mhm. Yes. So, that was >> There's also supposed to be a speaker in that arm.

[00:31:18]>> Speaker, >> something that looks like a speaker.

[00:31:21]>> I think that was a ventilation like a fan.

[00:31:23]>> Oh, ventilation. Okay.

[00:31:25]>> Definitely need a fan for that many actuators in one place.

[00:31:29]>> Right. It makes noise >> like a speaker.

[00:31:32]>> Oh, 100%.

[00:31:35]I mean uh so my my biggest comment to be honest was about the the cover design because um okay starting a bit from early there was a claim yesterday which was the what was it the big the best hand in the history or something right?

[00:31:48]Yeah. In history. Yeah.

[00:31:50]>> So, uh that type of claims they're they're very much flawed when it comes to robotics. Mostly because you see the best thing in the history of something and not when it's getting published, but after a few years of it being published.

[00:32:05]And that's specifically for robotics, specifically for Dexter's hand is important because all of these things that we see here, they will show if they are good or bad after a while, after hundreds of thousands of cycles. And with none of the hands we we can actually prove or disprove that. So again that that type of claims I don't really like but there are so many positive things about this hand compared to all the other hands we've seen. Uh to me that the covering was interesting. Uh one of the reasons was that for a while the industry was so much in loved with fabric covering and also like situational silicon paddings. But then they they managed to put everything into one silicon glove which is not easy at all.

[00:32:52]It's extremely hard. I've been through that in some projects and it's not really possible. And I also saw in one of their clips the one that is uh taking off grapes if you have that one. Um >> this one >> playing with grapes. Yeah. There you can see that the side of the hand is actually broken.

[00:33:10]>> Yes.

[00:33:12]If the >> I think the gray part was in the beginning of this clip or something.

[00:33:16]>> Yeah. Yeah. So >> yeah, exactly.

[00:33:20]>> On the right side.

[00:33:22]>> Yeah, you're right, Mir Dad. Uh when we talk about these coverings, I think we have to be a bit cleareyed about u the the pros and cons. And you know, the the penetration or let's say the the tear that you end up seeing in there, whether it's a fabric glove or that usually comes through because they're trying to give a they're trying to eliminate the pinch points, right? Because it seems like everyone's knuckles has a pinch point. So like, well, if you just put a covering over it will eliminate the pinch point. The problem is the pinching is still there. It ends up pinching the material and the result of the pinching the material after enough cycles is it wears out and breaks. So you've got to do something fundamentally different to even make sure that your coverings uh do not do not fail like that. And >> very interesting point absolutely and also like as you said the pros and cons for example one of the cons of fabric in hand was always the hygiene factor >> and among other things and but this system could actually solve the hygiene issue very well by just being I don't know washable or cleanable or something like that and I think this is a very very fair advantage. However, uh one thing that was a bit surprised uh there were two things actually. One is the the sensors. How do we uh how does the touch touch sensors can detect through this uh very overall glove or something and they haven't >> I can I can answer that. I can answer that. Have you ever put on a latex glove like when you're cleaning the dishes?

[00:34:49]>> What what happens to your senses?

[00:34:52]>> Good question.

[00:34:53]>> I've never done that. I don't wear >> Oh, you don't do the dishes? I wear I wear gloves. I mean, I actually do.

[00:34:58]>> Yeah. They get dullled, right? You know, that's why we don't like wearing gloves when we're trying to do something.

[00:35:04]That's fine. It's just dull offenses.

[00:35:07]>> But when the surgeon puts on his gloves, >> it's horrible.

[00:35:12]>> Sometimes they don't, right? Sometimes surgeon don't do it with a bare hand.

[00:35:15]>> Yeah. I mean, I shouldn't mention names, but one of the best plastic surgeons in the world. I mean when he gets really frustrated and it's really fine he rips off his gloves and just to finish the the small sutures. I mean but that is not common.

[00:35:31]>> If you when we do like u when we fix bones with plates and screws and nails, we use double paired gloves to kind of protect ourselves and the patients in case there's a rip in the glove. But when you're doing micro surgery or finer work, it's not fun to have double gloves. So, but we can of course work with gloves. We can even have double gloves and still do a lot of >> and and that's why >> So, what does that tell you about the limitations of these bots of these hands >> like >> I think we talked about it a few weeks ago also when you have gloves you cannot have fingernails or if you have it that ext that's extremely uncanny to have fingernail like a solid part on top of your fingers on top of the glove.

[00:36:13]>> That that's one thing. The other thing is just on the tactile sensing. I know for a fact that you get so much noise on the sensors only because of this extra layer that is moving also on the palm of the hand. And also the third thing I want to mention was the camera that there was a time that all humanoid hands started to uh or integrate camera inside the palm or inside the wrist. But then at some point they all kind of uh moved back from it and uh same we see it here.

[00:36:45]I don't remember exactly if 1x had cameras in the beginning but uh that's >> I don't think they ever did. I I don't recall >> I think Yeah. Yeah. They they never had it in the hands.

[00:36:54]>> I don't I don't think they ever did. Uh Roden, I just sent you an image that gives you like a better closeup of of where one of the terrors is because I I wasn't seeing in the shots that we were making.

[00:37:04]And the other thing I wanted to mention here was that the the the Bowden uh Bowden tubes in the mechanical in the mechanical video they were moving quite a lot. Right? We could see them while the wrist is moving. They're also moving a lot. But in the final video, we don't really see the the cover or the the wrist moving at all. That that's my concern.

[00:37:28]>> That that movement go like >> it's going to go in an enclosure of some sort and that enclosure is going to put them in there and that's going to make those Bowen tubes kind of move a very different way.

[00:37:39]>> Exactly.

[00:37:40]>> And we know one >> restrict them then it will >> they put a pretty thick cover over their arm for protection. So they've already shown that before that usually it's like this lattice structure that's going over and I think they would be doing the same thing to give you the protection which means that thing is like really enclosed in there. So >> uh yeah they they need room to be able to move otherwise they're going to kink >> and I notic that same problem with the Tesla hand when is that also those tubes are all over the place which is why everyone gave up on the tube and just a heck let's just do a straight shot >> 100%. And also I really don't know if these tubes how rigid they are because if they are very like as rigid as a bicycle tube then they will actually put a lot of strain on the on the wrist actuators. So their movements affects the wrist actuator and vice versa. the wrist acture to movement will affect the finger but we have talked about it before but this is the reverse effect also that comes here >> right >> and so they should be very soft but if they're very soft how come they have a very big loop of or the turning loop >> yeah and and Chris you know you're seeing movement there and that movement means energy >> it doesn't come for free so there there's a certain amount of loss going through those So I don't know on the paneling either they made the panel much bigger than what we see here. So there is a bigger offset much bigger offset or there is something here for example on the final video. Maybe they shut down some of the degrees of freedom. I'm just guessing for example the the metacarpal that you said for here they I haven't really seen it in the final video activating. I don't know.

[00:39:27]>> It only moves when it's kind of standing like this out in the courtyard. You could see that it moves forwards and backwards and it's kind of down here.

[00:39:34]So, it at least moves, but I'm not sure if it's an active movement or if it's just like a passive flexion extension as the finger is moving. So, at least it is moving when it's out in the green area or whatever and flicking his fingers.

[00:39:49]Um, >> yeah, there here is still without the panel, right?

[00:39:54]So I'm thinking about when we have the glove, does it actually work or not?

[00:39:58]Because with the glove, that would also be very complex movements uh to get.

[00:40:04]>> Mhm.

[00:40:05]But they showed the video of the tactile sensors and in the text they at least write that they are incorporated into the skin that the tactile sensor kind of in the glove. I'm not sure that that's what they're meaning if it's in the glove or not.

[00:40:17]>> But so the reason I said that is because usually these tactile sensors, they have too many wires. They have small wires, but they should be so so many wires. I didn't see any wire coming off the the wrist. I know, Scott. Did you see any wire?

[00:40:31]>> I didn't see any either. I mean, we're just seeing the the routing of the the tendons, but yeah, I didn't see any any wires at all.

[00:40:40]>> For some reason, we can't see it. But they're showing when it's gripping like the the grape with the gloves. That video is really zoomed in for some reason.

[00:40:47]>> And then you can use just the gloves at least. It might be a mock but still mock up but it looks like it's actually using the gloves to >> I mean you can be tell operation still take a great like you can try to fix your precision so I don't know if you see enough proof here for tactile sensing to be honest like from my my understanding >> and all of these can be achieved with teleoperation with gap >> okay is is there a wait a minute I'm seeing another actuator up there right in the middle that I hadn't seen before.

[00:41:21]There seems to be a real big one. Is that the one for the wrist?

[00:41:24]>> It looks like it.

[00:41:25]>> Yeah.

[00:41:28]>> Tendon crossing though.

[00:41:29]>> So, is that the So, okay. So, that might be where the 11 comes from because we're seeing like 10 along there, right? Are we able to sort of count 10 actuators as we go down that stack?

[00:41:42]>> Yeah, defin minimum of 10. I think there is some outside like below where the screen >> I count 11 actually. Yeah.

[00:41:49]>> 11 is >> Yeah. And and again 11 times two that gets us 22 because technically a degree of freedom is like is is control is something you're able to move. It's not joints >> which is what we've always had everyone kind of confuse a number of joints with the degrees of freedom. So we've had people talk about you know like a 22 do hand where they're counting up all the joints but then it turns out there's really only six motors actuating it. Or usually it's like the 11 do hands usually have six motors. So they're really six degrees of freedom even though they have 11 degrees of movement.

[00:42:22]>> And and again those that's you could also be kind of splitting hairs there is like well you're pairing it on there so that's not really a true degree of freedom. You know it's like >> yeah kind of in one direction. So you you're >> I So I think that's where the count is really coming from is is the number attendance >> probably. Yeah.

[00:42:43]>> Yeah.

[00:42:45]And it has yeah what else is there? Oh yeah the 45 newtons for each finger I think is pretty good. The thumb is usually considered a little bit stronger but 45 newtons for the individual fingers I think is similar to human normal human thumb might be a bit stronger as I said in that power group.

[00:43:02]Um >> and I mean two million cycles do we know that? Have they have they had time to test it for more than >> Yeah. That that's what I was also wondering. Is that their >> Oh yes yes yes. As a matter of fact, yes. Because remember, uh, Barren uh, was posting back in the spring a bunch of things about, oh, we just hit like a million cycles and then, you know, I think a weekend later, oh, we just hit like two million cycles. So, I think they had something where they were going through and were running it. And again, they were probably running it u on like nominal load and not going out to what would be the maximum or yield strength.

[00:43:38]>> I mean, there are some nice clips. Some of them are, of course, they we we don't see everything, but they're like pulling up or untying a zipper or we call unzipping a zipper.

[00:43:48]>> And they have it's not Lego, so I'm not going to give them that. It's actually Duplo that they're putting together, so it's a lot easier, but still they they're doing the Lego test and kind of pushing pieces together.

[00:44:00]>> I also suggested peeling an orange. They also didn't do that.

[00:44:04]>> That is that is extremely hard. But >> and then >> Oh, now they're cherry. Oh, these are grapes. Okay.

[00:44:11]>> Yeah, they're not cherry picking things here.

[00:44:12]>> No, they're not cherry picking things, >> but they did the whole connecting of a cable. It's not a USB stick that we are looking for, but still connecting some kind of >> Mhm.

[00:44:24]>> This was also good because this is almost one of the few times that we see use of tool and use of tool we know that is almost the most important thing for Dexter's hand, right? use of human tools and this this thingy is being designed for human hand proportions. So it it shows a lot of success at least in this area.

[00:44:45]>> And yeah, >> so let me quickly um this is not really good. I mean the game it's fun but it's not really playing the controller.

[00:44:59]It's like >> I had a hard time to understand if it can reach actually the the top button.

[00:45:05]>> No, exactly.

[00:45:08]>> And also Scott Scott shared this image, right? Let me just bring it up. Scott, you wanted to make a point here.

[00:45:18]>> Oh, yeah. I mean, you can see there on the on the pinky finger, you can see that there's a hole >> right in the center. So, so that's an example of one of the tears. And you see that's happening around a point where you probably have an internal pinch point >> that's just going ahead and as it straightens out you know eventually it just is going to eat its way through.

[00:45:36]So, you know, the the irony here is that it's what uh IP68, >> which means, you know, water intrusion, stuff like that, which means no dust and everything else. And and usually um that means like no pinch points, you know, there's there's nothing you can get pinched on. But unfortunately, the the actual protection mechanism is being pinched >> and and this is this is the problem that they all have. Um it seems like you you you should be able to make something that when you move it, it just doesn't open up like that. you know, I've seen mechanisms like this before. Again, going back to Kyber, I think if you look at the way their the top of their IP joint is, they don't really expose anything that uh you consider a pinch point.

[00:46:18]>> True.

[00:46:20]>> Also, maybe that's it's because of the over what did you call it? Um the hand >> uh the hyperextension. You think >> extension? Yeah, because these covers are designed for one way uh flexion actually. So if it goes the other way, maybe that that would risk >> and it's even hyperextended flaps.

[00:46:39]>> Yeah.

[00:46:39]>> Yeah. Look at it right now. That pinky is hyperextended.

[00:46:42]>> That is correct. Yeah.

[00:46:43]>> The pip. Yeah.

[00:46:45]>> Yeah.

[00:46:46]>> And I mean that might be part of the tendon design. I mean I tried to build the deck sand a few years back and it had that risk kind of as a side effect.

[00:46:57]>> Put a hard stop in there. I mean, what I mean, you just have the mechanical hard stop that bam, it hits it and it can't go any further.

[00:47:03]>> Yeah. Okay.

[00:47:06]>> So, um, who still has their booking for the 1x Neo?

[00:47:10]>> I'm waiting.

[00:47:11]>> Oh, I do.

[00:47:12]>> I'm still waiting for my Cyber Truck and my 1X.

[00:47:17]>> Looks like a good deal if they can have one.

[00:47:20]>> I'm not worried about having to pay up to >> Yesterday I was in Makina Summit in Paris two days ago. Also, Burnt actually was there and a few other people. He didn't say anything about the hand, by the way, in his keynote. And then he went back home very quickly after his pitch and then at night we saw the announcement. Anyway, but >> Well, well, you you weren't evidently he did spill the beans to a couple of people there.

[00:47:45]>> Yes, apparently privately.

[00:47:47]>> Privately he did. He privately did.

[00:47:49]Yeah.

[00:47:49]>> Okay.

[00:47:50]>> It was And it made its way through the grapevine.

[00:47:54]>> Yeah. I don't know if but uh Jonathan Hurst from um agility I was talking to him and he told me that oh we are now working on our fifth generation which will be rebuilt in um November sometimes and it would be much bigger than this uh agility digit that we have and the legs are not backward actually the legs are humanlike very surprised with a few news interesting news there but yeah let's see let them uh publish it themselves anyway the the point about that was that we we had this discussion with a few people um including uh including um Mark from uh Boston Dynamics the the goat. And then um somebody asked, "Oh, if you want to bet on any of these human companies today, if you want to buy it now, put all your money in, which one would it be?" and almost everybody had the consensus that probably none and we need to wait for one or two more years to get a clearer answer. And then uh the people who were more on the investment side or the the money people, they were like, "Oh yeah, but that's too late if you if you want to wait for that long." But then the people on the technical side were like, "Well, yeah, we we we still cannot really prove or disprove anything with with what we see." So that was that was a take at least.

[00:49:18]>> Now I cannot really disclose all the names who were there but yeah it was it was a fair amount.

[00:49:23]>> So what are you saying? It's now time to jump in.

[00:49:27]>> Well as an investor you're risk taker right? But people in technology they are not really taking those big risks yet.

[00:49:33]So it seems like there are a lot of investors who are actually willing to take these risks but not uh >> Scott knows a few.

[00:49:43]Yeah, Scott went but Scott went in a very very interesting way for >> that's your cue for Robbo strategy to plug it.

[00:49:53]>> I not investment advice.

[00:49:57]Sorry if I'm a bad bad PR for robot strategy here, but >> no no no no no just no >> I think a good thing about robot strategy is that you have distributed uh pro portfolio right so you have multiple companies that one at least one of them will work but if you want to bet on one company I think many people would say by today that is crazy crazy >> it it is hard but at the same time I I agree with kind of your your analysis that um if you're investor like when you have the clear signals is then it's too late. You've you've lost the alpha, you lost the edge, right?

[00:50:29]>> The whole idea of venture capital is to try to get that. And sometimes you're right and sometimes you're wrong. And you really try your best um to not be wrong either way. I mean, it's like some of the investors you talk to out there and you ask what was their biggest mistake and you assume it's going to be like the company they should not have invested in and it's always like the one that got away, the one they should have, the one they looked at and said, "Oh, this, you know, I don't." But at the time it would have been the same thing that all the technical people would have said is like this is not a safe bet.

[00:50:59]It's not going to work.

[00:51:00]>> So yeah, that's that's the hardest thing about investing.

[00:51:04]>> And it's like everybody's trying >> everybody's trying to make that pitch for the same for the same uh kind of to the same VCs, right?

[00:51:14]>> So how do you differentiate when when all the tech guys are telling you the same thing, hyping it up?

[00:51:23]To be honest, I think the hype is not coming from the technical people, right?

[00:51:27]>> The hype is coming from somewhere else.

[00:51:29]The technical people are trying to be very realistic of the of the shortcomings.

[00:51:34]>> And then >> I mean to be to be fair, hang on. To be fair, and I'm not I'm not throwing shade at 1x, but there was a lot of criticism about what 1x was doing right for a while in the middle.

[00:51:48]hotel operating or >> Yeah, I mean they kind of there's a lot of people saying they kind of lost their way. There were people coming and going.

[00:51:58]Um but then they've pulled this out of the hat. Yeah. So they've kind of redeemed themselves. Would you would you say this is a new benchmark or is that too much the new industry benchmark for the hand?

[00:52:12]>> I mean it's too it's too little that we've seen. I mean there's a lot in this that we have seen from different companies. A lot of the the blog language is similar to what we're hearing from other groups as well. So I mean it's a really nice ham. I'm really impressed with what they're doing, >> but it's it's still kind of a a stage as all of these are. There are movies that they prepared. It's not putting it to the test in a live situation where I think figure it's only the the American companies done that. There are Chinese companies doing it as well. But that's where we really see when the the the silicon hits the road, the rubber hits the road. I mean, >> so >> I had to try this.

[00:52:53]>> Dang, what do you think? Uh what is the true test, the true litmus test that every VC should kind of look out for?

[00:53:03]>> I would say um for for investing in humanoids.

[00:53:07]>> Mhm.

[00:53:09]Um it depends on your your timeline, right?

[00:53:14]So like if some of them some of them are looking at medium-term like you would be looking at someone like Boston Dynamics or or Appronics who are hedging their bets with other kind of robotics, industrial robotics, right? And I've seen this a lot in China where they do have a good humanoid program like Limix, but they also have a very solid robot which is not a humanoid but it earns money right away.

[00:53:43]>> Yeah.

[00:53:44]>> And feeds the cash flow to fund the humanoid.

[00:53:47]>> So yeah, these five figure hands and legged humanoids are kind of longer term and you want to take a bet longer term also.

[00:53:57]um from that perspective.

[00:53:58]>> Very true.

[00:53:59]>> But for for the hand in particular, I think like somebody said like we want to see demos where the all the all the fingers have been used because a lot of the demos we see is like packet sorting or whatever. We see just two fingers and and the argument is the grippers can do it. Why use such complex mechanism and an expensive hand?

[00:54:22]>> So yeah.

[00:54:23]>> Yeah. In fact, um, by the way, we should have a separate chat about, uh, your, uh, amazing trip to China. I would love to get the lowdown from you. Uh, all of us would, in fact. Um, so we should set that up. But yeah, but that's a that's a really important, um, point you make, Devang, because guys, we've we've been talking so often here on the channel about different form factors and different use cases, right? It's like, is the industry kind of getting carried away by focusing just on human rights?

[00:54:53]It's not focusing just on humanoids.

[00:54:56]>> Yeah, >> that's where a lot of the attention is going there, but there's definitely plenty of other companies that are coming up with bmanual kind of robots.

[00:55:04]They just don't get maybe as much attention, >> but they maybe are actually getting more commercial traction and so you're not hearing about it, but you know, they're being used in a lot of different companies or a lot of companies are already running proof of concepts for many of them.

[00:55:20]>> Yeah. I mean there was this study I don't know if it was recently published but at least it was taken into the into the Twitter vers or experts about this surgical bot that they had where they had like a unitary bot doing laparoscop laparoscopy. Oh yeah, >> as a proof of principle, but it was fully terop operated as far as I could tell. And when you do this, it's like it's basically grippers that you're using. It's not actual humanoid hand or humanoid form. You're teleoperating a robotic, not to put it down. I haven't read the paper. It's probably a lot of good findings there, but it's similar to just putting a robotic head on on a Dainci bot or something and saying, "Oh, look. It's a humanoid." But I mean it is probably gaining traction and might help you get funding and driving your research. So and I'm still convinced that the the the MVP most valuable player will be the humanoid. It's not going to be uh one bot for each different task. When you when you solve the humanoid, you will have the the the golden ticket. I I agree with Gustav in that sense also because again as we have said before so many times like single application robots doesn't make sense economically and has so many shortcomings but one other thing I wanted to add is that when we um we have talked also before that the the blockage here is data collection and training it's really not on the hardware side but mostly on the software side and training and it's believed and it's proven that training a humanoid with a human form factor Dexter's hand is much more straightforward than training a system that is different from human.

[00:57:05]>> Yes, absolutely.

[00:57:06]>> I think this is what you're talking about.

[00:57:08]>> Exactly. I mean >> there is a >> there is a remote operated >> um operating robot called Davinci, isn't there?

[00:57:18]>> Yeah. Yeah. And I mean this is like a version of that. They mentioned it in the same paper that that is the golden standard but the future will be having humanoids doing surgery according to Elon in three or I guess two and a half years now believe it when I see it but I mean this is nothing different than the Dainci aside that it's fewer arms working and it's teleoperated by surgeon sitting there and they show you how they kind of prepare everything with a human in the loop. So sure it can kind of >> but there's not a human on the table.

[00:57:50]>> It's a pig or something.

[00:57:51]>> It's an animal. So it's at least a live demo in a way.

[00:57:54]>> Okay. I thought it would be a kadaavo or something though.

[00:57:56]>> It might be a kadaabra. I'm not sure, but it's another human.

[00:58:01]>> I I wonder if it has anywhere near the same accuracy and feedback that the da Vinci has, >> especially with this distance that it has with the end point of the >> contact.

[00:58:13]>> Yeah. And that that would amplify whatever backlash the the motors might have.

[00:58:19]>> I mean, I guess this is we will see a lot more of this in the coming years to kind of build up hype and you need to test it. Maybe it's a bit early, I'd say, to kind of even the proof of principle. It's too early. They're not really showing a humanoid, I'd say. At least not from this clip. As I said, I haven't read the paper. So, you could just hypothesize that this will work.

[00:58:42]that would give as much information I'd say as doing it this way. But that's just >> actually I have another take from what Gustaf said before and it's that I think it might be very wrong saying that but I think we need the hype >> in order to be able to raise money.

[00:59:00]>> We need the hype to be able to raise money for the R&D of these companies because again we all know on the technical side that you're not there yet but if we want to get there we need this amount of capital and this amount of time. So the hype helps to raise enough money to be able to put it in R&D and and do the work. If there was no hype, basically it was just the money was coming with like just dropping uh with small amounts and that was that that that couldn't really move the industry so much forward.

[00:59:30]>> Let me ask you this. Um how far along the R&D road do you think the industry is? Is it I mean because you some would say you're just starting out some would say getting very close to deploying you even though it's with very useful hands.

[00:59:48]>> My take is that first of all we are building on top of 35 years of research that is already been done right. So it's not definitely zero it's not even also 10% probably much higher than that maybe up to 50% or higher. The issue is that especially when it comes to dextrous manipulation, a lot of the researchers in this area claim that there are a lot of unknowns. So we don't know where is the end of the road. So therefore we cannot say how far we are from there.

[01:00:15]And those unknowns are basically areas that for example when you said the the optimum form factor have we receive achieved the optimum form factor yet?

[01:00:25]Definitely not. Uh or something that everybody have consensus on. What what is the reason? because we don't really know what is the end of the road, what what other possibilities we could have in the coming months and years. So therefore, I think there there is no really an accurate answer to that question, at least from my my point, >> Edique.

[01:00:47]>> All right. Um let's get some closing thoughts from you gentlemen. Uh let's go around the panel. Um let's start off with you, Scott. final take on what we're seeing and um what does the industry have or competitors have to do now to follow or match intriguing hand. Uh it may let's say set a couple of new benchmarks. Um but I don't think I would say it is the most impressive hand in history. Uh, to me that still goes to the shadow hand doing the Rubik's cube.

[01:01:28]And and when I see this thing doing the Rubik's cube, then I'll say, "Okay, you're in kind of the same league as that."

[01:01:34]>> This hand did do the Lego blocks, though. The Lego blocks, which is >> two blocks.

[01:01:41]>> Okay. Um, Devon.

[01:01:45]Yeah, definitely one of the most impressive hands but in general the industrial industry is still in very early innings for the hands. Uh the body and hands are not not in sync with uh the tech maturity yet and uh we have long ways to go from seeing the cost coming down. I'm I'm going to be uh touring another company proception tomorrow. check out their hand, which is also very impressive.

[01:02:17]>> And uh yeah, it's just an exciting field to follow.

[01:02:21]>> Good stuff.

[01:02:23]>> I mean, I also like the demo. It's a really nice hand design. It's similar to what I expected them to be. Oh, here you have the really freaky hyper. Um I mean, it seems like they're listening to some of the inputs that are going around. So we're not we can't take credit for everything we took credit for as a test point of view talking to a hand surgeon but I mean there are glimpse of what we have been talking about and it might just be that we are thinking in the same direction and having similar ideas and that they are doing the same homework and I mean yeah it's a big step forward I'm just happy that we got a nice hand demo there seems to be a few new ones coming out now so I'm still waiting for Gen 3 Optimus But maybe this will give me a kick in the behind >> and mar.

[01:03:15]>> Yes. Uh in addition to what all the gentlemen said. Uh so I think um there were a lot of good improvements uh such as the the coverage. I really liked it.

[01:03:25]It's extremely hard to get such coverage but they they managed it. and also small form factor human like we've seen these from other companies but I think here they show all the the whole package together uh was was basically an advantage um I expected more on the perception side and on the sensor side uh to see more on that area but couldn't really see and also I disagree with their claim that this could be the this is the best hand in the human history that that that is still very to me >> oh my gosh Whoa. Feisty. Spicy. I like that.

[01:04:05]>> Push back. Finally.

[01:04:07]>> Fire me.

[01:04:09]>> Yeah.

[01:04:12]>> Well, all right. That's a great deal to end up. Thank you so much, gentlemen. So great to catch up at very short notice and um shout out to Barren and team.

[01:04:21]Great job. We're looking forward to a lot more from you guys. So, till next time.