特斯拉Optimus对决Figure 3：人形机器人大比拼

所以两个人形机器人走进了你的房子。其中一个是特斯拉的Optimus机器人。它有点像空手道小子那样展示肌肉，觉得自己很厉害，想当个格斗家。

So two humanoid robots, they walk into your house. One of them is Tesla's Optimus robot. It kind of flexes like a karate kid. It thinks it's really tough. It wants to be a fighter.

然后还有Figure three，来自今天刚宣布的公司Figure。这个新机器人只是安静地装载洗碗机，去叠毛巾，然后站到一个巨大的无线充电器上给自己充电。所以问题来了，哪个才是真正接近实用的助手？你更想要哪个在你家里？

And then there's Figure three, which is from a company named Figure that just was announced today. And this new robot, it just quietly loads the dishwasher. It goes and folds the towels, and then it goes and recharges by standing on this giant wireless phone charger. So it does the question too is, which one's actually closer to the real helper? Which one do you want in your house?

我们刚宣布了Figure three，我们会深入探讨它。然后还有特斯拉的Optimus机器人，我认为它显然是世界上人形机器人领域最大的竞争对手，在争夺最佳机器人的竞赛中。

We have Figure three just announced. We're gonna dig into that. And then we have Tesla Optimus robot, which is very clearly the, I'd say, biggest competition in the world of humanoid robots in the race to become the best one.

幸好我已经搬到了一个更安全的社区，Josh。所以我现在会选择那个人形助手机器人。我不常叠衣服，所以很想要个东西来代替我。说到这个，基本上就是Figure three，它是Figure公司新发布的机器人。我会形容它是一个超逼真的人形机器人，能做很多家务，Josh。

Thankfully, I have moved to a safer neighborhood since back in the day, Josh. So I I think right now, I would choose the humanoid helper robot. I'm not one for folding my clothes that often, so I would love something to replace it. Speaking of, that is essentially Figure three, which is the new robot released by the company Figure. The way I would describe it is it is a hyper realistic humanoid robot that does a lot of housework, Josh.

这之所以重要，是因为我觉得不仅是我，许多其他人也会使用它。我拿到了宣布视频，它 literally 不到一小时前才发布。所以Josh和我就像是刚拿到第一手消息。你可以看到它做很多事情，比如整理杂货。

And the reason why this is important is this is something that not just myself, but many other people I feel would use. I've got the announcement video, which literally just released less than an hour ago. So Josh and I are, like, fresh off the the press on this one. You can see it doing a bunch of things. It's sorting out groceries.

它在擦桌子，遛狗，扔球让狗去接，为你放毛巾。总之，如果给它戴上人类面具穿上人类衣服，除了可能有些机械动作，我可能会误以为它是人类。这很大程度上归功于他们的建造方式。

It's cleaning the table. It is taking the dog for a walk, throwing a ball for the dog to go catch. It is placing towels for you. All in all, I if if if you'd put this in a human mask and human clothing, I would mistake it for a human apart from maybe some robotic movements. And that's kind of like a lot of credit to how they built this.

但用Figure自己的话说，这个目标是提供一个真正通用的机器人，一个能执行类人任务并直接从人类那里学习的机器人。为实现这一愿景，我们的工程和设计团队完成了从零开始的硬件和软件重新设计。所以Josh，他们是从头开始建造的。这对他们来说是激情的劳动。我会说这个新机器人有四个主要亮点，让它如此特别。

But in Figure's own words, the goal with this was to deliver a truly general purpose robot, one that can perform human like tasks and learn directly from people. To realize this vision, our engineering and design teams completed a ground up hardware and software redesign. So Josh, they they built this from the scratch. This was a labor of passion for them. And I would say that there are four main highlights from this new robot, which makes it so special.

第一个是名为Helix的东西，这是他们的视觉语言动作模型。如果我们想想最流行的AI模型ChatGPT，它是一个LLM，即大型语言模型，对吧？你输入文本和字符，就会得到文本和字符输出。但对于机器人来说，它们不能仅凭文本操作。

Number one is something called Helix, which is their vision language action model. So if we think about ChatGPT, which is the most popular AI model, this is an LLM, a large language model. Right? So you input text and characters and you get text and characters out. Now for robots, they can't operate just from text.

它们需要使用各种其他输入，如视觉、音频和许多不同的东西，并将其转化为运动、视觉和声音。因此这是一个更复杂的问题需要解决，因此模型的构建方式也大不相同。Helix是Figure三的模型，是他们的专有秘密资源。他们所做的是训练这个模型直接与这个机器人协同工作，或者更确切地说，硬件是专门围绕这个模型设计的。

They need to use various other inputs like vision, audio, and a number of different things and translate that into movement, into sight, into sound. So a much more complex problem to solve, and the model therefore is built very differently. Helix is Figure three's model, and it's their proprietary secret source. What they've done is they've trained this model to work directly with this robot or rather the other way around. The hardware is engineered around this model specifically.

Josh，他们从这个模型和新机器人中强调的最酷的事情之一是视觉能力。相机架构相当惊人。它提供了比前代机器人高一倍的帧率，延迟只有四分之一，这意味着它反应超级快速和直观，视野宽了60%，这意味着它能看到更多东西。从安全角度来看更好，并且在适应你的生活方面也强得多。所以它可以用眼角余光注意到Josh正在离开房间，也许它会走到沙发旁为你整理靠垫。

One of the coolest things they highlight from this model and this new robot is the vision capabilities, Josh. So the camera architecture is pretty insane. It delivers twice the frame rate than its previous robot, one quarter of the latency, which means that it reacts super quickly and intuitively, and 60% wider field of view, which means that it just sees things. It's better from a safety perspective, and it's way better in terms of, like, acclimatizing to your life. So it could catch out of the corner of the eye that Josh is exiting the room and maybe he goes over to the sofa and tidies up the cushions for you.

对吧？它只是更加高度警觉。Josh，另一件事是手。我知道提这个有点奇怪，但你见过这个的照片吗？

Right? It's just way more hyper aware. The other thing, Josh, is the hand. And I know that's a really weird thing to to raise, but have you have you seen a picture of this?

是的。对于不了解的人来说，手是人形机器人最重要的部分。它们是最难创造和模仿的东西，而Figure有一个他们称之为触觉传感的新尝试。对吧？这就是我们在这里看到的，内置了摄像头？

Yeah. Hands, for people that don't know, they're the most important parts of humanoid robots. They are the most difficult thing to create and emulate, and Figure has a new attempt that they're calling tactile sensing. Right? Is this this is what we're looking at here with the This the cameras built in?

是的，是的。所以你在屏幕上看到的，对于只听音频的听众来说，这是一只超逼真的机器人手，但它的手掌中有一个摄像头。你可能会想，既然你已经告诉我它头上已经有像眼睛一样的摄像头了，为什么还要这样做？原因正如Josh提到的，手可以说是人形机器人最重要的部分。

Yep. Yep. So so what you're seeing on your screen, and for those of you who are just listening, is this hyper realistic robotic hand, but it has a camera in the palm of it. And you might be wondering, well, why the hell does it do that if you've just told me about a camera that it already uses in its head like like its eyes? The reasoning for this is, as Josh mentioned, the hand is arguably the most important part of humanoid robots.

对吧？它将搬运你心爱的东西，为你整理，你不想让它打碎玻璃杯。你不想它在房子里乱撞。因此，有了这个摄像头，它为这些微小动作提供了更好的感官深度，比如捡起一块布与捡起一个盒子，使用多少力，用多大的握力。这就是这个摄像头增强的功能。

Right? It's gonna be carrying the things that you love, tidying up after you you don't want it to break a glass. You don't want it to go smashing about the house. And so with this camera, it gives it a much better sensory depth for these tiny movements, for picking up a cloth versus picking up a box, how much force to use, how much grip to use. And that's what this camera is enhanced by.

正如他们描述的那样，这确实是一个全新的传感器套件，乔什。

And it's really a completely new sensory suite as they describe it, Josh.

是的。有几点。首先，对于硬件设备来说，通常第三代才是真正成熟的时候。我认为这次可能也是如此。第一版更像是测试版，是试验版本。

Yeah. A couple of things. One, with hardware devices, generally version three is when they actually get good. And I think that's probably the case with this. The first version is very much the beta version, the testing version.

能出厂就不错了。第二版修复了第一版的所有问题。那简直是一场灾难。然后第三代通过迭代学到了足够多的经验，基本上理解了前几代的缺陷，能够打造出一个真正好用、可大规模推广的产品。我认为我们今天在图灵身上看到的就是这种情况，他们已经发展到了第三代。

You're just happy to get something out of the factory. The second version fixes all the problems with the first version. That was a disaster. And then the third version has learned enough through iteration that it kind of understands the downfalls of previous generations and is able to build an actually good product that's usable, that's distributed at scale. And I think that's the case that we're seeing with Figure today, where they've made it through their third version.

他们经历了战斗的洗礼。衬衫上贴满了补丁，然后他们说，好吧。这是个好机器人。这是个能工作的机器人。这是个能为人们生活增添价值的机器人，而且它确实正在这么做。

They have gotten their battle scars. They have the patches on their shirt, and they're like, okay. This this is a good robot. This is a robot that will work. This is a robot that will add value to people's lives, and it actively is.

今天早上我读到的一个有趣的事实是，这个机器人实际上驻扎在一家，我想是宝马工厂或奔驰，一家汽车经销商里。对。它实际上已经在工厂里工作了几个月？是的。它确实在那里工作、做事，成为一个有生产力的机器人。

One of the fun facts that I got this morning while I was reading about this is this robot's actually stationed inside of a, I think, BMW factory or Mercedes, one of the car Yep. Dealerships. And it's actually in a factory months? Yeah. It's actually in there working and doing things and being a productive robot.

所以这很酷。其次就是手部。你提到了手部。它们里面有摄像头。起初，我对这个摄像头相当反感，因为就像全自动驾驶汽车中的激光雷达讨论一样，感觉添加这些不需要的传感器既愚蠢又多余。

So that's a cool thing. I think second to that is is the hands. You mentioned the hands. They have a camera in them. And initially, I was a pretty big hater on the camera because similar to the LiDAR conversation when it comes to full self driving cars, it feels dumb and redundant to add these sensors that you don't need.

就像，作为人类，我们手上没有摄像头。我们只有两只眼睛，用这两只眼睛，我们就能做所有需要做的事。所以你到底为什么要花额外的钱和技术在手部安装额外的传感器呢？然后我就想，你知道吗？如果我手上有个摄像头，我也不会讨厌它。

Like, it just as a human, we don't have cameras in our hands. We just have we got two eyes, and with those two eyes, we could do everything that we need. So why on earth would you spend extra money and technology on putting extra sensors in the hands? And then I was like, you know what? If I could have a camera in my hand, I wouldn't hate it.

就像，能够绕过物体去触及。你能看到平时看不到的东西。如果你能把眼球装在双手上。所以我觉得我开始改变对这些机器人增强方式的看法。意思是，我不讨厌你增加几个额外的传感器，就像全自动驾驶有点浪费一样，但对于人形机器人来说并不完全是浪费。

Like, could kinda reach around things. You could see things that you wouldn't normally see. If you could just stick an eyeball to both hands. So I think I'm starting to change my mind on on how these robots are augmented. Meaning, I don't hate if you add a couple extra sensors in the same way that it's kind of a waste for full self driving, but it's not quite for rope humanoid robots.

还有一件事，在你展示的那个视频里，它有触觉指尖，指尖对人形机器人来说也是非常困难的部分，因为你需要能够感知东西才能让它工作。比如，如果你的指尖没有感觉，你就不能确定抓握易碎物品时用了多大力。Figure公司做的是增加了这些触觉指尖。我们现在就把它们放到屏幕上。是的。

And another thing is in that video you were showing, it has, like, the tactile finger fingertips, and the fingertips are a really difficult part of humanoid robots too because you kinda need to be able to feel things in order for it to work. Like, if you don't have senses in the tips of your fingers, you're not really sure how hard you're grabbing onto fragile things. And what Figure did with this is they added these tactile fingertips. We'll put them on screen right now. Yeah.

在这些指尖内部，它们能够检测到低至三克的力，这是非常轻的。所以你可以触摸一些相当轻的东西。我不会太担心它触碰易碎物品，这真的很重要。我们还没提到的一个很酷的事情，我其实很喜欢的是，它通过脚部进行无线感应充电，这是相当新颖的。所以这个机器人会在你家里四处走动，做家务，然后只需站在这小小的充电脚垫上，通过感应充电自行充电。

And within those fingertips, they're able to detect up to three grams of force, which is pretty light. So you can touch some pretty light things. I wouldn't feel too worried about it touching fragile objects, and that's a really big deal. One of the cool things that we haven't mentioned yet that I actually loved is the fact that it has wireless inductive charging through the feet, which is something pretty new and novel. So this robot will kind of go around your house, will do the chores, and it will just stand on top of this little charging foot pad and charge itself up through inductive charging.

在充电的同时，它会以每秒10千兆比特的速度通过毫米波天线传输数据，上传到MegaMind并进一步训练这些模型。所以这真是一个很酷的迭代。我的意思是，这个版本，我记得他们说比Figure Two轻了9%。它的运行时间标称约为五小时，行走速度约为每秒1.2米，这足以作为家庭内的初代版本。我认为他们在这个演示中想展示的就是，嘿。

And while it does that, it will dump off data at 10 gigabits per second on its millimeter wave like antennas to upload to the the MegaMind and further train these models. So it's a really it's a cool iteration. I mean, this version, I think they said it's 9% lighter than figure two. Its runtime is listed for about five hours, and it's good for about 1.2 meters per second in terms of walking speed, which is sufficient enough to be a version one inside of a house. And I think that's kind of what they wanted to showcase in this demo is, hey.

这个机器人现在已经足够好用了。就像，你真的可以把它放在家庭里，它会提供某种价值。价值有多大？我不知道。清理一个人类三十秒就能搞定的沙发，它会不会花十分钟？

The the robot's good enough now that it works. Like, you can actually put this in a household and it will serve some sort of value. How much value? I don't know. Is it gonna take ten minutes to clean up a couch that would take a human thirty seconds?

很可能。但它会在你不在场、不去想它的时候完成。就像我认为人们对Roomba机器人非常适应，那些只是扫描地板并进行清洁的，像拖把和吸尘器之类的。我认为人们会很快开始适应这个。Gijo，有一件事我没听到，就是价格。

Probably. But it's gonna do it while you're not there and while you're not thinking about it. And the same way I think people are very comfortable with, like, the Roomba robots, the ones that just kind of, like, scan through your floor and clean it up, like mops and vacuums and stuff. I think people will very quickly start to get comfortable with this. The one thing that I didn't hear about was pricing, Gijo.

所以我真的很想知道他们实际将这些产品推向市场的策略是什么。你有没有找到关于定价或一般市场进入策略的信息？

So I'd love to know what the what their strategy is for actually taking these to market. Do you find anything on on pricing or or just general go to market strategy?

不。这是最简单的答案。

No. It's the simple answer.

是的。所以可能还没有定价，但他们正在做的是建造一个专门的工厂来冲压这些东西。我认为他们之前迭代版本的问题在于，他们是用定制工具来制造所有这些部件。他们做了定制工具，用 CNC 铣床来制造这些东西。

Yeah. So there there's probably no pricing yet, but what they are doing is they're building a dedicated plant just to stamping these things out. I think what they did with the previous iterations that really just didn't get the job done is they were custom tooling all of these things. They made, like, custom tooling. They were creating these things using CNC mills.

他们花了很长时间，成本也很高。他们建了一个新工厂，称之为 BotQ，专注于压铸、注塑和冲压，这些东西非常具有可扩展性，制造起来也很容易。

They it took a long time. It was very expensive. They've built a new factory. They call it BotQ, which is all about die casting and injection molding and stamping, and these things are are very scalable. They're very easy to make.

它们大幅降低了成本。所以虽然没有谈到成本，但他们谈到了产量，初期目标是每年生产 12,000 台，然后在接下来的四年内达到 10 万台，这看起来是合理的。我认为很多时候，这些机器人最难的是制造。我们见过很多看起来很棒的机器人演示，但没人能一次制造超过 10 台，因为他们没有相应的制造能力。所以建一个年产 12,000 台的工厂已经很了不起了。

They lower the cost quite a bit. So while they didn't talk about cost, they did talk about output, which is hopefully 12,000 of these per year initially, and then a 100,000 over the next four years, which seems reasonable. I think a lot of times with these robots, the hardest thing is to make them. Like, we've seen so many sick robot demos that look incredible, but no one can make more than, like, 10 at a time because they don't have the manufacturing for it. So building a factory that is good for 12,000 units a year, that's great.

这目前是领先的。我知道特斯拉为 Optimus 机器人制定了新的路线图，他们计划以数量级的方式生产更多。但解决制造问题，或者说至少正在解决制造问题，是机器人领域最困难的部分。所以看到 Figure 团队在这方面取得进展，非常非常令人乐观。

That is a leader so far. I know Tesla has a new road map for the Optimus robot. They plan on making probably a lot more by an order of magnitude, But having solved manufacturing or at least being on the way to solving manufacturing is the most difficult thing about robots. So very, very optimistic to see that covered by the Figure team.

是的。我认为非常重要的一点是强调这个机器人是完全自主的。没有远程操作员在某个外国通过摄像头观察它的动作并在出错时改进或帮助它。这个东西就在你家里，通过脚趾或脚部或其他方式自行充电，然后自主地完成各种任务。

Yeah. I think it's really important to emphasize that this robot is completely autonomous. There is no teleoperator somewhere in a foreign country that's kind of, like, looking through the camera and seeing the actions that it does and improving it or helping it when it goes wrong. This thing sits in your home. It charges itself via its toes or feet or whatever it might be, and it just does things autonomously.

它还会主动从你的习惯中学习。所以它不是某种预编程的机器人。它通过摄像头观察一切，聆听声音，每天都在进步，最终成为你理想的家用机器人。但问题是，乔什。

It learns actively from your habits as well. So it's not some kind of preprogrammed robot. It's looking at everything through its camera feed. It's hearing things, and it's improving each and every day to become, you know, your ultimate household robot. But here's the thing, Josh.

实际上，它不仅仅是一个家用机器人。在这个视频的结尾，我要快进一下，你会看到它在酒店担任礼宾员。我得说，这简直是有史以来最酷的礼宾员。如果我走进一家酒店，有这家伙告诉我该去哪儿，我基本上会惊叹不已。这是他们试图表达，或者说Figure公司试图表明：嘿，我们不仅锁定了酒店服务业，还涵盖了体力劳动这一块。

It isn't just a house robot, actually. At the end of this video, and I'm gonna skip ahead here, you see it as a concierge at a hotel. I mean, I have to say that's the the coolest concierge ever. If I walk into a hotel and I had this guy telling me where to go, I would be like in awe essentially. And is their attempt to say or Figure's attempt to say, hey, we've got the hospitality industry on lockdown as well, but we also have the manual labor side of things.

对吧？所以你看到其中一个Figure机器人在递送包裹。可以想象，未来的亚马逊配送或任何类型的递送服务都将由这些机器人管理。我当然希望它们更便宜、更精准，而且不会损坏我的包裹。FedEx，我说的就是你。

Right? So you're seeing, one of the figure robots delivering a package. So you can imagine Amazon deliveries or any kind of deliveries in the future being managed by your robots. I sure as heck hope that they are cheaper, that they are more precise, and that they don't damage my package. You know, FedEx, I'm talking about you.

你还可以看到这个机器人的工厂版本。Josh，你之前提到，我认为宝马正在用它来制造汽车。现在，这样一个机器人可以帮你分拣包裹。你可能觉得这没什么了不起，但亚马逊曾花费超过15亿美元试图用他们自己的手动机器人来实现自动化。所以这并非易事。

You're looking at a factory version of this robot as well. Josh, you mentioned earlier that, I think BMW is using it to construct cars. Now one of these robots can sort your packages out. And you might not think that that is cool, but that was a problem that Amazon spent upwards of, I think, 1,500,000,000.0 to try and automate with their own manual robots. So it is not an easy task.

这是一项极其昂贵的任务。我相信有很多工厂会乐意使用这些东西。所以关键是，Figure 3虽然被推销为家用机器人，而且我认为这是一个很好的策略，也是大多数人会购买的原因，但我认为实际上有一个秘密买家在工业硬件领域，他们会说：嘿，我想要一双便宜、不会坏、还能为我全天候运行的手。我觉得这太棒了。

It is an incredibly costly task. And I'm sure that there are a number of different factories that would gladly use these things. So the point is Figure three, whilst it's being pitched to retail as this home robot, and whilst I think that's a really good approach and it's why the majority of people will buy these things, I actually think there's a secret purchaser, which is in the industrial hardware world where they're like, hey. You know, I would love to use a pair of hands that are cheap and that won't break down and that can run $24.07 for me. I think that's great.

最后我想提的是，我不想低估你关于规模化这些机器人的观点，Josh。因为如你所说，100个机器人影响不大，但1亿个机器人就开始发挥作用了。Brett在这个发布视频和他的博客文章中的评论都围绕着一个核心：专门为规模化建造这个机器人。你提到的制造这些机器人的工厂是从第一版就完全重新设计的，目的就在于此。所以我很兴奋，想知道我们是否最终都会在家里拥有一个。

And then the last thing I wanna mention is I don't wanna underscore the point you made, Josh, around scaling these things. Because as you said, like, a 100 robots doesn't really make an impact, but a 100,000,000 robots now are cooking. And Brett's commentary over here in this announcement video and in his announcement blog post was all around building this robot specifically to scale. The manufacturing plant that you mentioned that builds these robots was completely redesigned from the first version of this robot with this in mind. And so I'm excited to see whether we all end up with one of these in our homes.

是的。制造业真的……我无法形容这有多难，以及它对我们讨论的这些机器人公司的成功构成了多大的限制因素。我真的很尊重和钦佩Figure对此非常重视，似乎比几乎所有其他公司都更认真，除了特斯拉。10万个是一个严肃的数字。我的意思是，那将是很多人形机器人在四处走动。

Yeah. Manufacturing is is really like, I can't understate how difficult that is and how much of a, like, limiting factor that is in the success of these robot companies that we're talking about. I really respect and admire the fact the Figure is taking this very seriously, seemingly more so than than nearly every other company outside of, I would say, Tesla. A 100,000 is it's a serious amount. I mean, that is a lot of humanoid robots to be walking around.

目标就是你所说的那个数字，Ejaz。比如，我们能否达到1亿个？他们能接近1亿吗？因为随着这些东西变得更有价值、更实用，它们开始解决更多用例，产生更多生产力，这可以抵消经济的重大部分。如果它们能进入工厂，如果它们能处理通常由家政服务员做的家务。

The goal is going to be that number you said, Ejaz. Like, can we get a 100,000,000 of these? How close can they get to a 100,000,000? Because as these things get a little bit more valuable, more practical, they start solving more use cases, they start generating more productivity, that can offset significant parts of the economy. If you can go into factories, if they can take care of things at the house that a normal housemaid would do.

这些事情确实非常重要，必须做好，因为如果做对了，对人类劳动力的需求就有无限的上升空间。如果一个机器人能完成人类的工作，却没有人类的那些缺点，那将是一件大事。我认为目前Figure在这场竞赛中处于领先地位。我不确定是否见过有机器人能如此出色并具备规模化能力。

These things are are really important to get right because if you get them right, there is an infinite upside in terms of demand for human labor. And if a robot can do human labor without any of the human downsides, that's a really big deal. And I think right now, Figure sits at the top of this race. I'm not sure I've seen any robots that are this good and have the capability of scaling.

别太自信了，Josh。你忘了排名第一的公司。第一名哦，

Your horses, Josh. You're forgetting about the number one company. The number one Oh,

我绝不会忘。

I would never.

它它它的名字和felon husk押韵。我我我有个很酷的视频，在我们深入讨论之前，Josh，虽然不是完整的人形机器人，但它是人形机器人的一部分。是一只手的视频。你能解释一下这里发生了什么吗？

It it it rhymes with felon husk. I I I have a cool video before we get into it, Josh, which is not an entire humanoid, but part of a humanoid. It's a hand. Can you can you explain what's going on here?

好的。回到手这个话题。手是机器人最重要的部分。它们——我们的手出奇地复杂。就像，如果你现在坐在这里听，不妨试着张开手。

Yes. So going back to the hand thing. Hands, most important part of a robot. They are the it's surprisingly complex, our hands. Like, if you just kinda if you're sitting here listening, just kinda flare up.

没错。所以你的手有这些叫做自由度的东西。网上很多人称之为DOF，自由度指的是一个关节系统能独立运动的方式数量。人类的手有27个自由度。我们的四根手指每根有四个，拇指有五个，手腕有六个。

Yeah. So your hands have these things called degrees of freedom. A lot of people online call them DOF, and a degree of freedom is how many independent ways a joint system can move. The human hand, we have 27. So each one of our four fingers has four of them, our thumb has five, and our wrist has six.

所以我们的手能做的所有这些动作——横向移动、弯曲、折叠——都是我们的自由度。你在特斯拉Optimus演示中看到的是，这只手现在有22个自由度，而之前只有11个。所以22是我们最接近人类手的程度，这非常重要，因为一旦达到27，并且一旦具备了人类手的灵巧性以及正确的感官输入和输出，机器人基本上就拥有了人类与世界互动的全部能力。因为我们的手能做27种不同的动作。如果机器人也能做到，理论上，我们的能力和人类机器人之间就没有区别了。

So all of these things that our hands can do, move laterally, bend, folds, those are all our degrees of freedom. What you're seeing here in the Tesla Optimus demo is that this hand now has 22 degrees of freedom and the previous was just 11. So 22 is the closest we've gotten to the human hand and that's really important because once we reach 27 and once you get the dexterity of a human hand with the correct sensory inputs and outputs, you essentially have the full capability that a human does in terms of how you engage with the world. Because our hands, we can do 27 different movements. If a robot can do the same, in theory, there is no difference between our capabilities and a human robot.

我们现在看到的是，Optimus已经非常接近了，而这将是他们新版本中即将推出的手部，我们还没见过。所以我很兴奋，敬请期待。但Figure也一直在应对这个挑战。虽然我有点失望，因为他们没有具体说明Figure三号手有多少自由度，看起来数量不少，我们可以参考之前的手部。Figure二号，也就是上一代机器人，他们的手有16个自由度，这相当不错，因为Optimus第二代只有11个自由度。

And what we're seeing here is, I mean, Optimus got pretty freaking close, and this is the hand that's gonna roll out in their new version that we haven't seen yet. So I'm very excited since staying tuned for that. But Figure has also been on this challenge. And while I'm a little disappointed, I don't think they've specified how many degrees of freedom the Figure three hand has, it looks like it is a good amount, and we can go based off of the previous hand. So Figure two, the previous robot, their hands had 16 degrees of freedom, which is pretty good because Optimus Gen two had 11 degrees freedom.

所以在自由度方面，他们遥遥领先。Figure三号，我们还不清楚。看起来很不错，但也许这是个好机会来谈谈特斯拉在做什么，快速比较一下特斯拉和Figure。所以

So they were very much ahead in terms of degrees of freedom. Figure three, we don't know. They look pretty good, but maybe this is a good opportunity to jump into Tesla and what they're doing and just kinda quickly compare Tesla versus Figure. So

所以，是的。好吧。Josh，我们这里讨论的是不同的世界，具体来说，世界的另一边。这周早些时候，这个视频刷爆了我的推送，简直是最诡异的事情，因为我当时想，等等，这个机器人能打架。

So Yeah. Okay. So what what we're talking about different worlds here, Josh, be specific, the other side of the world. This video was just all over my feed earlier this week, and it was the most bizarre thing ever because I was like, wait. This this robot can fight.

我们现在屏幕上看到的是特斯拉的Optimus Prime机器人在和一位武术老师进行武术训练，而且出奇地好。我从头开始播放。它和那个人击掌，然后进入一整套动作序列。Josh，我首先注意到的是它的敏捷性。看它蹦跳的样子。

So what we're seeing on our screen right now is Tesla's Optimus Prime robot doing a martial arts session with a martial arts teacher, and it is surprisingly good. I'm gonna start it from the the beginning here. It's it's dapping the guy up, and then it goes into a entire sequence. And I guess the first thing that pops out to me here, Josh, is how agile it is. Look at it bouncing.

看它的手部动作，但也看其他所有部分的移动，它的头、它的腿。动作如此流畅。它在某个点绊了一下，然后自我纠正。我知道我们花了很多时间讨论手部方面，但看来埃隆一直在酝酿一种完全不同的机器人，一种相当实用的类型。我想这里要表达的观点是，我们这里有两种营销努力在进行。

Look at its hand movements, but also look at every other bit of movement, its head, its legs. It's so fluid. It stumbles at some point, and then it corrects itself. I know we've spent a lot of time talking about the the hand side of things, but it seems like Elon's been cooking up an entirely different type of robot, one which is pretty actionable. And I guess the point that's being made here is we have two types of marketing efforts going on here.

以经典的，我想，机器人风格，有人会说，嘿，你知道，这是一个生产型机器人。它可以替代体力劳动。它可以在你的工厂工作，但最重要的是，它可以在家里帮助你。它可以帮助你止赎。

In classic, I guess, robot style, you have someone that's like, hey. You know, this is a productive robot. It can replace manual labor. It can work in your factory, but most importantly, it can help you in your home. It can help you foreclose.

对吧？而埃隆则以他经典的风格介入，说，我要通过打斗序列来展示我的机器人有多敏捷和多能干。对吧？而且这还不是本周唯一疯传的视频。还有这个，我觉得这个超级酷。

Right? And Elon goes in classic Elon style and says, I'm gonna demonstrate how agile and competent my robot is, but through a fight sequence. Right? And and this wasn't the only video that was going viral this week. It was also this one where, I thought this was super cool.

他在《创：战神》电影首映礼上展示了擎天柱，基本上是在和现场的一位演员比身高，我猜是在挑战他打架吧。他还穿着打扮得像一个创战纪机器人，超级酷。我不知道，乔什。

He he premiered Optimus Prime at the Tron Ares movie premiere, where he's basically sizing up to one of the actors here and, I don't know, challenging him to a fight, guess. He he's also dressed and looks like a a Tron bot, which is super cool. I don't know, Josh.

你你还有

Do you do you have

什么想法吗？我觉得这超级酷。

any thoughts on this? I I thought it was super cool.

是的。首先，我们不能叫它擎天柱。擎天柱是变形金刚的名字。这个只是Optimus。

Yeah. First, we can't call Optimus Prime. Optimus Prime is the transformer. This one's just Optimus.

我就叫它擎天柱。我知道它不是擎天柱，但我觉得就是。

I call it Optimus Prime. I know it's not Optimus Prime, but it is.

好吧。所以这个机器人，我的意思是，这又回到了我们之前讨论的引言。就像是，你是想要这种非常冷静、酷炫、有条不紊慢慢完成任务的东西，还是想要这种可能行动迅速、打破常规的自然怪胎？我觉得他们是从两个不同的角度来处理的。Optimus机器人的速度和灵巧性非常出色。

Okay. So this robot I mean, again, it's this gets back to the intro that we were talking about. It's like, do you want this, like, very calm, cool, collective thing that's gonna slowly get the job done, or do you want this, like, freak of nature that's just going to, like, just maybe move fast and break things? I think they're kind of approaching it from two separate things. The speed and dexterity of the Optimus robot is remarkable.

它这么快就能如此迅速地移动，嗯。真是太令人印象深刻了。起初，我以为你展示的第一个视频是远程操作的，因为Optimus与现实世界互动的方式很多都是通过远程操作员。基本上，他们是人类，会在手上佩戴传感器，或者可能使用身体动作追踪服，然后基本上手动移动和控制机器人。所以机器人是在做动作，但AI并没有控制它。

The fact that it's moving this quickly, this soon Mhmm. Is so it's so impressive. At first, I thought that first video you were showing was teleoperated because a lot of the way that Optimus has been engaging with the real world has been through teleoperators. Basically, they are is they're humans that will wear sensors on their hands or maybe they'll use a body motion tracking suit, and they will move and control the robot basically manually. So the robot's doing the movements, but the AI isn't controlling it.

是人类在控制它。这个演示最吸引人的地方在于，是人工智能在控制它。没有人为输入，没有远程操控。机器人只是移动得那么快。

It is the human that's controlling it. What was fascinating about this demo is that's the AI controlling it. There is no human input. There is no teleoperation. So the robot's just moving that fast.

这怎么可能呢？深入研究特斯拉的供应链后，我开始意识到，它之所以运行得如此出色，是因为没有其他人制造这个机器人使用的执行器或电机。执行器是关节之间产生运动的方式。特斯拉内部制造了所有这些部件，因此他们实现了非常紧密的垂直整合循环，使得一切运转更快、效率更高，并且以更高的加速度提升速度。回到制造方面，特斯拉已经拥有工厂和规模化机器人的优势。如果你把Optimus想象成一辆车，它非常相似，只不过是把轮子换成了四肢。

Now how is this possible? And just going down the rabbit hole of Tesla's supply chain, I started to realize that the reason this is working so well is because no one else makes the actuators or the motors that this robot's using. The actuators being the way that the movement happens between joints. Tesla's building all this stuff in house, and as a result, they're getting this really tight vertical integration loop where things are just moving faster, they're more efficient, they're getting faster at a higher accelerated rate, And going back to manufacturing, Tesla already has that advantage of having factories and having scaled basically robots. If you think of Optimus like a car, it's very similar except you swap out the wheels for some limbs.

它使用相同类型的机器学习模型。我们在整个特斯拉技术栈中再次看到的是这种垂直整合。现在，他们正在进行用于汽车的AI训练，并将其与人形机器人训练相结合，所有这些数据都朝着一个单一使命前进。几周前我们有一期节目讨论了AI五芯片，情况也是如此。

It uses the same type of machine learning model. And what we're seeing again throughout the entire Tesla stack is this vertical integration. So now they're having their AI training that happens for cars. They're merging that with the humanoid robot training, and now all this data is going towards one singular mission. And the same thing we had an episode a few weeks ago with the AI five chip.

AI五芯片在汽车中运行，但现在它也将运行在Optimus上。因此，Figure仍然需要解决的许多制造问题，特斯拉已经通过汽车解决了，并且现在能够更快地大规模推出。我认为我们在这里看到的这个机器人不会是最终形态。它不会是我们家里的那个东西。据说，版本三将在今年晚些时候推出，就像我们之前提到的，版本三是好东西。

Well, the AI five chip runs in cars, but now it's also gonna run-in Optimus. So a lot of the manufacturing problems that Figure still has to figure out, Tesla has solved through the cars and is now able to roll it out at scale much quicker. I think this robot that we're seeing here is not not going to be a final form. It's not gonna be the thing that's in our house. Supposedly, version three is coming a little bit later this year, and like we mentioned earlier, version threes are the good ones.

我们想购买机器人的版本三。我们已经看到了Figure的版本三，我非常期待看到Optimus的版本三，因为它看起来很棒。我相信特斯拉作为一家公司能够大规模制造这些机器人。Figure目前感觉很像这个领域的Rivian，他们是明确的第二名，有机会做一些独特和新颖的事情。

We wanna buy version threes of robots. So we've seen Figure three's version three. We've seen Figure's version three, and I'm very much looking forward to seeing Optimus's version three because it looks great. I trust Tesla as a company to manufacture these at scale. Figure very much feels like the Rivian of the space right now where they are a clear second place with a chance to do something unique and novel.

我认为这些早期阶段是他们真正有机会脱颖而出的时刻，我认为两家公司都做得很好。我对两家公司都持乐观态度。我很希望能有机会与其中任何一家公司的人交谈，这样我们可以 firsthand 听到Figure计划如何击败特斯拉，因为他们面临着一场非常艰难的上坡战。

I think these early days are when they really have a chance to to break free, and I think both companies are doing great. I'm bullish on both. I would love to get the opportunity to chat with someone from from either of the companies so we can hear firsthand how Figure plans to to kick Tesla's ass because they got a very steep uphill battle coming.

看，我不确定我是否同意Figure在这里是第二名。我明白你在制造方面的问题确实很难解决，但似乎Brett和他的团队花了大量时间试图解决这个问题。而且我知道最近他们筹集了几十亿美元，天知道他们现在的估值是多少，但听起来他们也有资本储备来投入并解决这个问题。我相信他们有一批专家在为此工作，但我对最终目标的看法可能略有不同。虽然制造确实非常重要，你希望让一大批机器人投入使用。

See, I I don't know if I agree that Figure's in second place here. I I I get your point around the manufacturing side being a really tricky one to resolve, but it seems like Brett and the team have spent an inordinate amount of time trying to figure this out. And I know recently, like, they raised a couple billion, and I god knows what they're valued at right now, but it sounds like they also have capital arsenal to be able to spend and and figure this problem out. I'm sure they have a a bunch of experts to work on that, but I I kind of frame the end goal maybe slightly differently. Whilst manufacturing is gonna be really important, you wanna get a bunch of robots out there.

我认为第二个层面的影响，也就是他们从这些不同机器人获取的数据，将成为真正的游戏规则改变者。对吧？因为正如我们之前提到的，要训练这些视觉语言动作模型，你需要海量数据，而这些数据要么目前并不存在，要么极难获取。想想所有那些重要的体力劳动视觉数据。

I think the second order effect, which is the data that they get from these different robots is gonna be the real game changer. Right? Because as we mentioned earlier, like, to train these visual language action models, you need a ton of data, which either doesn't really exist right now or is super hard to access. Right? Think about like all this important manual labor vision data.

比如，工厂里工作的工人不一定佩戴摄像头，但现在一些机器人公司正在付费让工人们佩戴这些摄像头，以便最终取代他们。我看到很多在中国发生这种情况的视频。所以我认为这将是一场竞赛，看谁能获取最全面的数据、训练出最佳模型，并将其转化为任何类型的硬件组件。因此，我的一个大胆预测可能是：某些硬件组件最终会变得极其商品化。我认为埃隆会是第一个做到的，因为他正在内部自主研发，但它是否最便宜、是否最具可扩展性还有待观察，对吧？

Like, the the people working at factories aren't necessarily wearing cameras, but now some of these robot companies are paying workers out there to wear these cameras so that they could eventually be replaced. I saw a bunch of videos from this happening in China. So I think this is gonna be a race towards who can get the most comprehensive data and train the best model and translate that into whatever type of hardware component. And so I guess maybe a hot take would be some of these hardware components will eventually become incredibly commoditized. I think Elon is going to be the first one there because he's building it in house, but whether it'll be the cheapest, whether it'll be the most scalable is to be determined, right?

就像我们会尝试建立制造工厂和供应链，让Brett Adcock等世界各地的从业者能够更快地接入，单独购买执行器并加速前进。我不确定，但观看这一过程将会非常令人兴奋。

Like we'll try to be able to create a manufacturing plant and supply chain, which is just so much quicker for the Brett Adcocks and figures of the world to be able to tap in and buy the actuator separately and race ahead. I don't know, but it's gonna be pretty exciting to watch.

是的，有道理。从技术上讲他们确实是第一，但这有点像在Gemini 3下周就要发布的情况下，还说GPT 5是最佳模型。就像是，好吧，随便你怎么说。

Yeah. That's fair. I guess on the technicality they are in first place, it's kinda like saying GPT five is the best model knowing Gemini three is coming out next week. It's like, okay. Sure.

就像，你是最棒的。好好享受你的

Like, you're the best. Enjoy your time in

高光时刻，但是，系好安全带，因为大佬们

the spotlight, but, like, buckle up because the the big boys

要来了。这非常符合特斯拉现在的感觉。听着，我祝他们双方都好运。我超级兴奋我们正在迎来机器人时代，未来开始看起来像未来了。

are coming. And that's very much what it feels like for Tesla. Listen. I wish both of them the best of luck. I am so stoked that we are getting robots that like, the future is starting to look like the future.

这真是太酷了。嗯。

This is the coolest thing. Mhmm.

现在只剩下一个非常短暂的时间窗口，你还能走在街上而不会遇到人形机器人，或者你还能去处理你一天的差事。你不会遇到一个与你互动的人形机器人。无论是在你家里、加油站，还是在杂货店，

Is there is a very small window of time left where you can walk down the street, and there will not be a humanoid or you can go into your your errands for the day. There won't be a humanoid robot that you're interacting with. Whether it's in your home, whether it's at a gas station, whether it's at a grocery store,

这些机器人

these robots

正在穿过门。

are going through door.

走在我家附近的大街上。

Walking on my high street.

一个人形机器人？

A humanoid?

是的。一个人形机器人。它叫Rizbot，到处走动，让人们惊叹不已。我不知道你是否看过一些流传的视频，但我觉得它非常有趣。所以我不确定。

Yeah. A a humanoid. It was called the Rizbot, and it was walking around, whizzing people up. I don't know if you've seen some of the videos go by, but I I found it hilarious. So I don't know.

虽然还没到大众普及的程度，但我确实见过一个。

It's not mass appeal just yet, but I I saw one.

是的。这是一个非常激动人心的时刻。我为Figure感到兴奋，也为特斯拉和团队感到兴奋。两家公司都将推出非凡的机器人。

Yeah. This is this is a very exciting time. I'm stoked for Figure. I'm stoked for Tesla and the team. Both are going to have phenomenal robots.

我对此深信不疑。我衷心希望很快就能真正试用这些机器人。我渴望有个机器人四处走动打扫卫生，或者就在日常生活中看到它们。因为对于那些没听节目、不关注技术前沿的人来说，他们根本不知道发生了什么。他们完全没意识到我们正处在AGI即将到来的临界点。

I'm sure of it. And I am hopeful that the day will come very soon that we'll actually be able to try these. I'd love to have a robot walking around cleaning stuff up or just just see them in our day to day life. Because for the people who aren't listening to the show, who aren't really on the cutting edge of technology, they don't understand what the hell is happening. They have no idea that we are on the cusp of, like, we have AGI coming.

我们有了行为酷似人类的机器人。所有这些生产力技术将彻底颠覆他们的日常生活。我认为只有当机器人开始走上街头，只有当机器人出现在他们眼前时，他们才会惊呼：天啊，等等，这到底是怎么回事？

We have robots that act just like humans. We have all of this productive technology that is going to upset so much of their, like, day to day lives. And I think only once the robots start walking down the street, only once the robots are, like, in their face will it feel like, oh my god. Wait a second. What is happening here?

Figure在实现这一愿景中扮演着非常重要的角色。所以对机器人保持乐观的感觉很好。

And Figure is playing a very large role in making that a reality. So It feels good, feeling optimistic about the robots.

感觉我们快要迎来机器人领域的ChatGPT时刻了——嗯。它们确实相当出色，看起来近乎神奇，能为生活增添巨大价值。看看特斯拉这四张图：有做家务的，

It feels like we're getting super close to the chat GPT moment for robots where Mhmm. They're actually quite good. They actually seem quite magical, and they add a lot of value to our life. I mean, like, look look at these four pictures from Tesla here. You've got it doing housework.

有干户外杂活的，有在实验室捣鼓什么的（天知道在做什么），甚至还有在桑拿房里的——说实话，我不明白为什么需要看桑拿机器人待在桑拿房里。

You've got it doing outside chore work. You've got it at the lab cooking up something. God knows what. And you've got it in the sauna for some reason. I don't know why I would wanna see a sauna robot in the sauna, to be honest.

但我相信很多听众会想，好吧，我见过很多演示了。你们谈论了很多产品发布，但没有交付时间。所有这些特斯拉Optimus机器人目前还没有大规模量产。所以我们短时间内还无法在家里看到它。我现在能买什么？

But I'm sure a lot of you listening to this is like, well, okay, I've seen a bunch of demos. You guys have spoken about a bunch of launches, but there's no delivery time. All these Tesla Optima robots are still somewhat not at scale yet. So we can't see it in our homes anytime soon. What can I buy right now?

嗯，我有答案给你。不幸的是，展示它的视频非常暴力，但这是宇树科技的最新机器人。我认为是宇树三型，你现在可以在沃尔玛以21,600美元的价格买到，乔希。我以为这是假的，所以我自己上了沃尔玛网站，不幸又幸运地报告这是真的。不过，我不建议虐待你的机器人。

Well, I have an answer for you. And unfortunately, the video that's displaying it is a very violent video, but it is UniTree's newest robot. I think it's UniTree three, which you can now buy from Walmart, Josh, for the casual price of $21,600. I thought this was fake, so I went on the Walmart website myself and unfortunately fortunately can report that it is real. I I do not advise abusing your robot, though.

我看到的很有趣。这些公司采取了不同的路线。宇树科技非常注重先打造一个低成本产品并解决营销问题。所以我在YouTube视频博客上看到宇树，在Instagram上看到人们讨论它。

What what I've seen it's funny. There's different tracks that these companies are taking. Unitree is very much taking the build a low cost one and solve for the marketing first. So I see Unitree on, like, YouTube vlogs. I see people talking about it on Instagram.

它确实

It does

也能做后空翻。

these backflips too.

非常令人印象深刻。非常令人印象深刻。我认为它是一个相对愚蠢的机器人，我不确定它能做到我们在演示中早些时候看到的Figure机器人展示的事情，但它是一个非常酷的机器人。就像Sora二型因其新奇性而令人惊叹一样，宇树因其新奇性而不可思议。比如，大多数人从未见过人形机器人四处走动，更不用说能像这样站起来做后空翻的了。

It's very impressive. It's very impressive. I think it's a dumb robot, relatively speaking, where I'm not sure it can do the things that we saw Figure demo earlier in the the presentation, but it's a very cool robot. And if like, the same way that Sora two is amazing for the novelty, unitary is incredible for the novelty. Like, most people have never seen a humanoid robot walking around, let alone one doing a backflip that can stand up like this.

如果你不是看着波士顿动力公司的视频长大的，你以前从未见过这个。所以这真是一件非常不可思议的事情，如果你能用21,000美元得到它，那就是某人的营销预算。你只需把这个坏小子贴上你的标志，带出去，制作几个视频，在营销方面你就会大获成功。所以，是的，不同的方法。比如，宇树绝对是目前唯一能买到的。

And if you didn't grow up watching Boston Dynamics videos, you have never seen this before. So it's like this really incredible thing, and if you can get that for $21,000, that's someone's marketing budget. And you just slap your logo on this bad boy, you take it outside, you create a couple of videos, you are going to hit a home run terms of marketing. So, yeah, different approaches. Like, unitary is definitely the the only one that you could buy right now.

在智能方面它可能是最差的，但我的意思是，它太酷了。看看你能对它做什么。如果你只是想要一个摔跤的伙伴

It's probably the worst in terms of intelligence, but it's I mean, it's sick. Look at what you can do to it. If you just want someone to wrestle with

如果，如果Go能做任何事。听到这里并想买一个的话，请务必购买，因为我是不会买的。并且告诉我们你的体验如何，因为我非常想知道在家里拥有这样一个东西是什么感觉。最后严肃地说一句，我想强调Figr创始人兼CEO布雷特·阿德考克今天宣布Figr3时提到的一点，那就是没有人形机器人能在没有AI的情况下实现规模化。这才是关键所在。

If if Go do anything. Listen to this and are tempted to buy one, please buy one because I'm not. And let us know what your experience is like because I would love to know what it's like to have one of these things in your homes. Getting serious for for one final second, I wanna highlight something that Brett Adcock made, the founder and CEO of Figr when he announced Figr3 today, which is there is no way humanoid robots scale without AI. And that's what it comes down to.

如果AI赢了，机器人并不会赢，我认为推动这两个行业向前发展至关重要。我完全不认为马斯克先生如此执着于机器人和AI是巧合。他正在运营xAI。他刚从我们昨天讨论过的黄仁勋那里获得了20亿美元的投资。你一定要去看看那一期节目。

Robots don't win if AI wins, and I think it's super important to push both industries forward. I do not think it's a coincidence at all that mister Musk has been so diehard on robots and AI. He's he's running x AI. He got a 2 bill investment from Jensen Huang that we spoke about yesterday. You should definitely go check that episode out.

但总而言之，乔什，这些行业真的是紧密交织在一起的。看到两个行业同时蓬勃发展真是令人惊叹。你我不久前，大概是三个月前，在一期节目中说，我对每周都有新模型发布感到非常厌倦。但说实话，我们有点被宠坏了，因为能身处一个这些模型飞速进步的世界是一种殊荣。而现在我们看到这一切正在机器人技术领域扎根。

But all in all, these industries are really heavily intertwined with each other, Josh. And it's just amazing to see kind of both industries flourish at the same time. You and I, not too long ago, I think was about three months ago, said on an episode, I'm getting so tired of new model releases every week. But realistically, we're a bit spoiled because it's such a privilege to be in a world where these models are rapidly advancing at such a quick speed. And now we're seeing the roots of that happening in robotics.

所以总的来说，这真是一个激动人心的时代。

So just all in all, really exciting time.

是的。太神奇了。机器人技术完全是另一回事。对我来说，智能部分已经是个已解决的问题，因为有太多人在攻克它。困难的是，你能造出这些东西吗？

Yeah. It's amazing. It's robotics is a whole different animal. Like, to to me, the the intelligence part is a solved problem because so many people are tackling it. The difficult thing is is can you build these things?

就像，原子的世界和制造业比数字比特的世界要困难得多。智能固然重要，但对我来说那感觉像是个已解决的问题。真正让这一切成为现实的重任落在了这少数几家公司身上。这是一个非常艰巨的挑战，愿意承担的人要少得多。所以我们需要Figure。

Like, the world of atoms and manufacturing is so much harder than the digital world of bits. And the intelligence is important, but to me, that feels like a solved problem. It's on these very few companies to actually make this a reality. Like, this is a very difficult challenge that far fewer people are willing to take on. So we need Figure.

我们需要特斯拉，我们需要宇树科技。即便是这个，像是被踢来踢去的小东西。制造这些东西是最困难的部分，而且并非理所当然。所以，是的，真的非常乐观。

We need Tesla. We need Unitree. Even this, like, this little thing that's getting kicked around. Like, building these things is the hardest part, and it's not a given. So, yeah, really, really optimistic.

感谢布雷特·阿德考克。向你致敬，兄弟。你正在付出非常艰难的努力，打一场逆风仗

Grateful for for Brett Adcock. Shout out to you, dawg. You're putting in some really difficult effort fighting an uphill

战斗。布雷特。

battle. Brett.

是的。来上节目吧。请来上节目吧。我们很想和你聊聊。但是，EJ，我想今天就到这里吧。

Yeah. Come on the show. Please come on the show. We'd love to talk to you. But, EJ, I think that wraps it up today.

就这些了吗？

Is that is that all we got?

是的。就这些了。

Yeah. That is all we got.

这是一期机器人主题的节目。

That's a robot episode.

各位。感谢大家加入我们，一起讨论我们最热衷的极客话题和最新爱好——机器人。乔什，每次我们做这类视频，这些机器人都变得越来越逼真，简直让人害怕，我已经等不及Optimus正式发布的那天了，我相信到时候我们肯定又会做一期评测。如果你喜欢这期节目，请点赞、订阅，并分享给你的朋友们。我们再次从大家那里收到的关于我们应该讨论的新话题或我们遗漏的内容的反馈都非常棒。

You all. Thank you all for joining us to talk about our number one geek and latest hobby, robots. These robots are getting scarily more realistic every time we do a video on this, Josh, and I cannot wait for the Optimus grand release where we'll, I'm sure, do yet another review on these things. If you enjoyed this episode, please like, subscribe, share it with your friends. The feedback we've been getting once again from you guys on new topics that we should talk about or things that we missed has been superb.

请继续保持。给我们发私信，在我们的YouTube视频下留言，我们会回复你的。我想今天就到这里吧。我们下期再见。祝好。

Keep it up. Give us DMs, comment on our YouTubes, and we will, respond to you. And I guess that's it for now. We will see you on the next one. Peace.