探访中国人形机器人AI训练实验室
Inside the AI labs training China’s humanoid robots
译文简介
难得，一篇《金融时报》的文章，正文除了最后一段有点儿阴阳怪气，其余的还算正常。
正文翻译
题图。
At a new 12,000 square metre facility in wuhan, young Chinese graduates spend their days piloting humanoid robots serving steamed buns, wiping down tables and folding laundry.
在武汉一处新建的1.2万平方米设施内，中国年轻毕业生们整天操控人形机器人，让它们端包子、擦桌子、叠衣服等。
Every movement the machines make inside the Rmb200mn ($29mn) laboratory’s mock kitchens and bedrooms is tracked and recorded by cameras and sensors.
这个造价2亿元人民币（2900万美元）的实验室里，机器人在模拟厨房和卧室中的每个动作都被摄像头与传感器追踪并记录。
The Hubei Humanoid Robot Innovation Center is one of dozens of state-funded robot training farms that have popped up across China to build a vast pool of robot-specific training data.
湖北省人形机器人创新中心是中国各地涌现的数十个国家资助的机器人训练农场之一，这些农场旨在构建海量机器人专用训练数据池。
“We’re like teachers and the robots are our students,” said Zhang Jia, a 21-year-old programme manager at the site. “When you teach a human, they get it after a few repetitions. But teaching a robot is different, you have to repeat actions hundreds, thousands, even tens of thousands of times.”
“我们如同教师，机器人则是我们的学生。”21岁的项目经理张佳表示，“教导人类时，重复数次便能掌握。但训练机器人就完全不一样了，每个动作需要重复数百、数千甚至数万次。”
Officials hope that amassing such data will help the country’s nascent humanoid robotics industry overcome key challenges facing AI as it moves beyond software into the physical world.
官员们希望，大规模积累此类数据将有助于中国新兴的人形机器人产业克服人工智能从软件领域迈向物理世界过程中面临的关键挑战。
插图：年轻毕业生在武汉一家培训实验室里教机器人拣货。
The push is part of President * **’s drive to make China the world’s foremost science and technology superpower. Last week, Bei*g identified “embodied intelligence” as one of six future industries to foster in its 2026-30 five-year plan, calling for the development of training centres, AI models and hardware to accelerate the rollout of humanoid robots.
这一举措是***推动中国成为世界顶尖科技强国战略的组成部分。上周，中国方面将“具身智能”列为2026-2030年五年规划中重点培育的六大未来产业之一，呼吁发展训练中心、人工智能模型及硬件，以加速人形机器人的部署和应用。
Experts say a shortage of robot-specific training data remains a major obstacle to translating recent advances in AI into real-world robotics. The large language models that underpin ChatGPT and DeepSeek have been fed vast amounts of text scraped from the internet. Robot data collection is still in its infancy.
专家指出，机器人专用训练数据的短缺，仍是把人工智能的最新进展应用到现实机器人中的主要障碍。支撑ChatGPT和DeepSeek的大型语言模型，其训练数据来自从互联网抓取的海量文本。而机器人数据的收集仍处于起步阶段。
Companies in China and the US are using a range of methods to gather training data, including real-world deployments, simulation and using AI-generated data.
中美两国的企业正采用多种方法收集训练数据，包括真实场景部署、模拟仿真以及利用人工智能生成的数据。
Tesla has explored using human demonstration videos to help train its Optimus robot, while Silicon Valley start-up 1X Technologies wants to put robots in households that will in part be remotely piloted by humans as they learn.
特斯拉已探索利用人类演示视频来帮助训练其Optimus机器人，而硅谷初创公司1X Technologies则计划将机器人部署到家庭中，让其在学习过程中部分由人类进行远程操控。
Beijing’s multiyear plan for the humanoid industry includes scaling robot training and data collection. Local governments from the rich coastal city of Hangzhou to the small interior city of Mianyang are pouring in money for new training farms. Hubei province, which holds the wuhan lab, has unveiled a Rmb10bn state fund for humanoids.
北京对人形机器人产业制定的多年规划包括扩大机器人训练和数据采集规模。从富裕的沿海城市杭州到内陆小城绵阳，各地政府都在为新训练农场投入资金。设有武汉实验室的湖北省已推出一项规模达100亿元的人形机器人产业基金。
“China is becoming smarter in how it supports emerging industries facing bottlenecks and that’s what they are doing with the data collection centres,” said Jay Huang, head of Asia industrial technologies at Bernstein research group.
伯恩斯坦研究集团亚洲工业科技主管黄杰表示：“中国在支持面临发展瓶颈的新兴产业方面正变得更加明智，这正是他们建设数据采集中心的意义所在。”
“Government support means the data is shared, benefiting everyone. It pushes everyone to work in the same direction,” he said.
他指出：“政府支持意味着数据实现共享，使各方受益。这推动所有人朝着同一方向努力。”
In wuhan, Zhang helps oversee 70 young instructors who work eight-hour shifts training their 46 robots. They use remote controls or sensor-equipped handheld devices to operate the machines, repeating the same movements over and over.
在武汉，张佳协助管理着70名年轻训练师，他们以八小时轮班制培训46台机器人。训练师们使用遥控器或配备传感器的手持设备操作机器，一遍又一遍地重复相同的动作。
插图：一位训练师正在教机器人如何清洁。
Nearby, rows of staffers review the video output, labelling every few seconds of footage with annotations such as “pivot left” or “extending arm”. The site produces about 100 hours of usable data a day, according to Zhang.
附近区域内，成排的工作人员正在检视视频画面，每隔数秒就对影像资料标注“向左转动”或“伸展手臂”等标签。据张佳介绍，该农场每天可生成约100小时的有效数据。
“We collect and organise the data, then upload it to our platform, where we label and process it,” he said. “We’re still in an exploratory phase.”
“我们收集并整理数据，然后上传至平台进行标注和处理，”他表示，“目前仍处于探索阶段。”
The idea is that the succession of sensor readings and videos that capture the positions, velocities and torques of the robot parts as they move can be fed into AI models for robots — typically “vision language action” models, in industry parlance.
其核心理念在于，通过一连串传感器读数和视频数据——这些数据捕捉了机器人部件运动时的位置、速度与扭矩——能够输入至机器人专用的人工智能模型，即业内所称的‘视觉-语言-行动’模型。
The process is intended to help recreate the breakthroughs seen with large language models in robotics. That could lead to machines learning more generalised skills, such as picking up a bottle of water without being explicitly programmed to do it.
这一流程希望在机器人领域复制大型语言模型那样的突破。未来或将使机器能够学习更通用的技能，例如无需专门编程即可自主完成‘拿起一瓶水’这类动作。
插图：机器在模拟厨房和卧室内的每一个动作都会被摄像头和传感器追踪和记录。
“Obtaining data at scale is not easy,” said Zhao xiang, a co-founder of Motphys, a start-up that has built an embodied intelligence simulation platform.
“大规模获取数据并非易事，”机器人初创企业Motphys联合创始人赵翔表示。该公司已构建了一个具身智能仿真平台。
Young engineers in their state-backed wuhan lab wear virtual reality headsets to train robots, allowing Motphys to collect data more efficiently and at a lower cost. “Simulation is essential,” Zhao explains. “An intelligence breakthrough may require hundreds of millions or even billions of hours of data.”
在这家获得政府支持的武汉实验室里，年轻工程师们佩戴虚拟现实头盔训练机器人，使Motphys公司能够以更低成本、更高效率采集数据。“仿真技术至关重要，”赵翔解释道，“一次智能突破可能需要数亿甚至数十亿小时的数据支撑。”
Experts said a more fundamental challenge loomed over China’s ambitions: for now, data collected from one robot cannot easily be used to power another with different hardware. Because the hardware is rapidly evolving, data collected now may not work as well for next year’s model.
专家指出，中国机器人产业面临一项更为根本的挑战：目前从一台机器人采集的数据难以直接应用于其他硬件结构不同的机器人。由于硬件技术快速迭代，当前收集的数据可能无法适配明年推出的新型号。
Bernstein’s Huang said data transferability remained an area of active research with progress expected. Google DeepMind’s robotics AI models, for example, have shown early promise in transferring certain skills across hardware platforms.
伯恩斯坦分析师黄杰表示，数据可迁移性仍是业界重点攻关领域，预计将取得进展。例如谷歌DeepMind的机器人人工智能模型已初步展现出在不同硬件平台间迁移特定技能的潜力。
China’s approach was laid out in a Ministry of Industry and Information Technology strategy document for the development of the humanoid robotics industry to the end of 2027.
中国的发展方针已体现在工业和信息化部印发的《人形机器人创新发展指导意见》中，该指导意见的有效期截至2027年底。
The plan identifies large-scale training databases and high-quality multimodal data as central to building the “brain” of humanoid robots. Local governments have rushed to support the programme, both to create jobs and seed future industries in their areas.
该文件将大规模训练数据库与高质量多模态数据确定为人形机器人“大脑”构建的核心要素。各地政府正积极支持这项计划，既为创造就业岗位，也为在本地区培育未来产业。
But even leading robotics researchers and the engineers carrying out the data drive said it remained unclear if it would yield the desired technological gains.
但即便是引领机器人研究的学者及参与这一数据收集行动的工程师也坦言，目前尚不确定这项数据工程能否产生预期的技术突破。
There is one tangible benefit: robot purchases by data collection centres have helped sustain China’s humanoid robot manufacturers while real-world demand for their hardware is still emerging.
但至少有一个明显好处：在硬件产品的实际市场需求尚未完全形成之际，数据采集中心对机器人的采购，为中国的人形机器人制造商提供了维持运营的支持。
The wuhan centre bought dozens of robots from Shanghai-based AgiBot at Rmb350,000 a piece. Bernstein analysts estimate sales for data collection made up about one-fifth of China’s more than 20,000 humanoid robot shipments last year.
武汉中心以每台35万元人民币的价格，从上海智元公司采购了数十台机器人。伯恩斯坦分析师估计，去年中国超过2万台的人形机器人出货量中，约有五分之一是用于数据采集。
At one data collection centre the FT visited, a dozen humanlike robots hung motionless on one side of a grand lobby.
在英国《金融时报》走访的一家数据采集中心，宏伟的大厅一侧十几台人形机器人被悬挂在那里，一动不动。
A staff member, who asked not to be named, explained that these robots were not being used for collecting training data. “These are the ones that perform when officials come to visit.”
一位不愿具名的员工解释说，这些机器人并非用于采集训练数据。“这些是供领导参观时进行演示用的。”
评论翻译
很赞 ( 5 )
收藏
Yet another example of how critically important state guidance and support is when pursuing national strategic obxtives, which, in this case, is to make China "the world’s foremost science and technology superpower."
这又是一个绝佳案例，充分说明了国家指导与支持在实现国家战略目标过程中的关键作用——在当前语境下，这个目标正是让中国成为“世界顶尖科技强国”。
@ping ping
The reason why China is winning in every technology vector has to do with scale. It’s all about scale. The number of engineers that graduated from Chinese universities over the past 2 decades dwarfs that of all of the west combined
中国能在所有科技赛道上领先，关键在于规模。一切都要归功于规模。过去二十年里，中国高校培养出的工程师数量，让整个西方世界加起来都相形见绌。
@Director of Communications
True. Scale is important. But even more important is having the capability to consistently set national strategic priorities and then to relentlessly pursue them until completion. And what about capitalism? I think it's critically important as the engine of growth, but capital shouldn't be the driver or conductor of that growth. This is where the West, in my view, has gotten itself into trouble.
确实，规模很重要。但更重要的是具备持续设定国家战略优先事项、并坚定不移地推进直至完成的能力。那么资本主义呢？我认为它作为经济增长的引擎至关重要，但资本不应当成为这种增长的驱动者或主导者。在我看来，这正是西方陷入困境的地方。
@Rkarete
Rent-seeking oligopolies pursue naked self-interest and starve the West of resources to pursue these policies.
寻租垄断集团只顾赤裸裸的私利，还让西方缺乏资源去推行这些政策。
As Bregman says, "Don't talk to me about your philanthropy, why don't you pay taxes"!
正如布雷格曼所言：“别跟我扯你的慈善事业，你怎么不缴税呢”！
@Director of Communications
Agreed. Here's an interesting quote from Zheng Yongnian, a professor at Chinese University of Hong Kong, taken from a NYT article: “The West’s decision to pursue neoliberal economic policies was a strategic mistake, which led to the de-industrialisation of their economies and provided China with an opportunity.”
同意。香港（特区）中文大学教授郑永年有段话说得挺有意思，这是《纽约时报》文章里引用的：“西方推行新自由主义经济政策是个战略失误，导致自身产业空心化，反倒给中国创造了机会。”
@Imperial Hookup v2
India has scale, but lacks the unified national drive and support to accomplish the stuff China does.
印度也有规模优势，但缺乏像中国那样举国上下的驱动力和支持，做不到中国所能做到的事情。
@JW
India seems to have the wrong type of poverty.
印度的贫困状况似乎走错了方向。
@Sisyphus Shrugged
Western Free Marketeerism will be remembered for the same dogmatic naivité that Soviet socialism is remembered for.
西方自由市场主义终将因其教条式的幼稚而被后人铭记，正如苏联社会主义也因其教条式的幼稚而被后人铭记一样。
@Contrarian Investor
Director of Communications - of ***? Valid point though.
你是中G的通讯主管？不过说得倒也在理。
@Director of Communications
Not at all. I'm a widely-travelled Westerner, who's lived in Asia for over 20 years. And in that time I've come to admire China and the Chinese. That's it.
完全不是。我是个游历甚广的西方人，在亚洲生活了二十多年。这些年来，我逐渐欣赏起中国和中国人。仅此而已。
And the fact that I'm a half-decent writer, an avid reader of politics and economics and a big-picture thinker, who, I suppose you could say, works at a central bank.
而且我勉强算个过得去的写作者，热衷于阅读政治经济内容，习惯从宏观视角思考问题——或许可以说，我在中央银行工作。
@Coyote
An army of robots trained in a lab in wuhan… What could possibly go wrong?
一支在武汉实验室里训练出来的机器人军团……还能出什么岔子呢？
@American Person
I suspect this will yield more fruitful results than messing around with Mars.
我怀疑这比折腾火星能带来更多实质性的成果。
@Apes
There is only one hospital planet for human life in the solar system and probably always will be (terraforming planets is a pipe dream) I’m not sure our aggressive ape like brains want to understand that .
太阳系里适合人类生存的星球只有一个，而且很可能永远如此（行星改造根本是白日做梦）。我不确定我们这些像好斗猿类般的大脑是否愿意接受这个事实。
@7Gf2K8mP
Who would have imagined that a strategic, long-term public system would outperform a government led by an aging reality star?
谁能想到，一个着眼长远的公共体系竟能胜过由过气真人秀明星领导的政府？
@By definition
And a 4-year elections cycle, split into mid-terms for good measure.
选举周期为四年，为了确保选举顺利进行，还会进行中期选举。
@guillaume esq
This article is excellent, a rare instance of reporting that goes beyond the hype to look at the actual way the sausage is made. The comment on the maturing way in which the Chinese government supports industry via the creation of a common pool of resources usable by all actors is very interesting, looks like industrial policy done right. A lot to ponder.
这篇文章非常出色，难得一见地没有被炒作蒙蔽，深入剖析了实际运作机制。文中关于中国政府通过建立可供所有参与者使用的公共资源池来支持产业发展的成熟方式的评论，非常有意思，这看起来是一次成功的产业政策实践。有很多值得深思的地方。
@NiZnaio
Why China wins.
这就是为什么中国能赢。
@H00dw1nk3d
Humanoids to address ageing problem.
人形机器人用来应对老龄化问题。
Renewables to address climate change and energy security.
可再生能源用来应对气候变化和能源安全问题。
Rare earths and advance manufacturing to claim technology supremacy and having a chokehold on the world.
稀土和先进制造业用来争夺技术霸权并卡住全世界的脖子。
I gotta say they have played their cards very well.
不得不说他们这手牌打得真漂亮。
@YoungReezy
They have, but other cards have included Borg-like oppression of their population and tech theft.
确实，但他们手里的其他牌还包括像《星际迷航》中博格人那样压迫人民以及技术盗窃。
@A
They do this by stealing , coercion, spying, cyber attacks, and industrial espionage. *** is tumor of the world
他们通过偷窃、胁迫、间谍活动、网络攻击和工业间谍来实现这一点。***是世界的毒瘤。
@Mervin K
China has hit peak population due to past policy. Their population will shrink and age. Normally it’s impossible to grow an economy with a falling population - but you can if you have next generation AI humanoid robots .
由于过去的政策，中国人口已达到峰值。人口将减少并老龄化。通常情况下，人口下降时经济几乎不可能增长——但如果你拥有下一代人工智能人形机器人，那就有可能实现。
Put in those terms we are witnessing a brilliant long term government strategy.
这么一说，我们正在见证一项卓越的长期政府战略。
@Digital-nomad
One that is not possible in our Neo-con society because you'll spook "the market".
在我们这种新保守主义社会里，这种事根本不可能，因为你会吓坏“市场”的。
@Chenicus531
China isn't so fussed about population anyway. 1.4 billion is way too much and the government I believe is perfectly happy to halve the population to 0.7 or even 0.5 billion in a generation. It cares about managing the transition, so robotics would be used heavily in household chores and carehomes.
中国其实对人口问题没那么焦虑。14亿人口实在太多了，我觉得政府巴不得一代人内就把人口减半到7亿甚至5亿。他们操心的是怎么平稳过渡，所以机器人技术会在家务和养老院领域被大量应用。
@Le Gun
Don’t worry, here in the UK we’re investing £10s of billions a year in lifelong not working families. Much better use of taxpayers resources.
别担心，我们英国每年都往那些终身不工作的家庭里投好几百亿英镑呢。这纳税人的钱花得可真值啊。
@Customgauge
This will likely see them way ahead of the US and Europe in the way they are with drones.
这很可能让他们在机器人领域的发展方式上，远远领先于美国和欧洲，就像他们在无人机领域的优势一样。
DJI drones are already irresistible because of their extremely high quality for consumers. America and Europe can make good drones, but only really with a military purpose in mind - there is no real attempt to go at the consumer market. This gives a huge advantage to China when collecting data via DJI, and therefore rapid development.
大疆无人机因其面向消费者的极高品质已经令人无法抗拒。美国和欧洲也能造出好的无人机，但基本上只着眼于军事用途——并没有真正尝试进军消费市场。这为中国通过大疆收集数据并实现快速发展提供了巨大优势。
When they create consumer humanoids to do household chores and sell them to us - and we will buy them because they will be cheap and better - the collection of data will broaden still more. They will hear every word, and know every task - even how many t shirts you own. Fascinating strategy.
当他们制造出用于家务劳动的消费级人形机器人并卖给我们时——我们肯定会买，因为它们会既便宜而且质量更好——数据的收集范围将进一步扩大。它们将听到每一句话，了解每一项任务——甚至知道你拥有多少件T恤。这策略真是耐人寻味。
@British Oligarch
Meanwhile in the UK. We have Labour government. That’s it. That’s the joke.
与此同时在英国，我们有个工党政府。就这样。这就是全部笑点了。
@Digital-nomad
And before that, we had a Tory government. And soon we'll have a Reform government. That's a sick joke.
在那之前，我们有过保守党政府。很快我们又会迎来改革党政府。这真是个恶心的笑话。
@Sagittarius Traveler
Apart from the aspects regarding Chinese research on humanoid robots, I was struck by the fact that in the laboratory building in wuhan, researchers appear dressed in parkas as if outside. The problem of heating homes and work or educational spaces in China remains a serious one.
除了关于中国人形机器人研究的那些方面，武汉实验楼里研究人员穿着像在户外一样的厚羽绒服工作的景象也让我很惊讶。中国在家庭、工作场所和教育空间的供暖问题依然严峻。
@Anonymous 0416
China is big and wuhan is considered as a southern city, not a northern one. That means, they have air conditioner in summer rather than heating in winter. But what you’re saying is correct, just for more information.
中国幅员辽阔，武汉通常被视作南方城市而非北方。这意味着当地夏季会使用空调，冬季却未必有集中供暖。不过您说的没错，我只是补充些背景信息。
@Contrarian Investor
Good spot. In Southern China, there is no central heating, and in some places, it can get very cold in winter. Northern China is much more comfortable.
说得很对。华南地区没有集中供暖，有些地方冬天能冷到骨子里。相比之下，北方冬天就舒服太多了。
@joining the dots
While American religious extremists are blowing up the Middle East, China is rising!
当美国的宗教极端分子在中东四处放火时，中国正在崛起！
@Cosmopolitan
The beginning of the end of humanity. Superintelligent and superstrong robots ruling at one point over us, going rogue etc. whatever. Just a question of whether it takes 100 years or 200 years. But what is the end goal here? I just see darkness eventually. But hey... sure we had some "productivity gains" and our "lives were made easier" for a couple of decades....
人类终结的开端。超级智能和超级强壮的机器人终将统治我们，失控什么的，随便啦。问题只是需要100年还是200年罢了。但终极目标到底是什么呢？我只看到最终一片黑暗。不过嘛……当然啦，我们确实享受了几十年的“生产力提升”和“生活便利”……
@AGIA
I don't think you have to wait for more then 10 years to see the darkness of it all.
我觉得用不着等上十几年，就能看清这一切有多黑暗了。
@Sisyphus Shrugged
And the sad part is that the killing machines won’t even be thinking killing machines, they’ll be rogue artificial parroting machines.
可悲的是，这些杀戮机器甚至不会有杀戮机器的自觉，它们只会是失控的人工复读机。
I’m increasingly suspecting that although LLM / GenAI and the like will NEVER bring us General Artificial Intelligence (ie the actual thing the term AI was initially thought for, everything else is marketing branding), LLM / GenAI can nevertheless bring our civilization existential risk.
我越来越怀疑，尽管大语言模型/生成式AI这类东西永远无法带来真正的人工智能（即最初构想中那个真正意义上的AI，其他都是营销话术），但它们仍可能给我们的文明带来存亡危机。
Future Alpha Centauri xeno-archeologists will be running digs on Earth only to conclude those degenerate creatures living inside caves originate from a race that invented LLM-based automated ICBM targeting software that inadvertently created a hidden deadman switch when its black box data used Civilization V streams on YouTube as its training data.
未来的半人马座外星考古学家发掘地球遗迹时，只会得出结论：那些穴居的退化生物所属的种族，发明了基于大语言模型的洲际导弹自动瞄准系统——该系统因使用《文明V》游戏在YouTube上的直播流作为训练数据，意外创建了隐藏的死手开关（注：即死手系统，确保在国家指挥机构被摧毁、无法人工下令时，系统能自动发起核打击，实现“确保相互毁灭”的终极威慑）。
@jagcx75
C's national strategic obxtive is: "Copy Elon". (EVs, rockets, humanoids)
C的国家战略目标是：抄袭马斯克”。（电动车、火箭、人形机器人）
Let's see what Optimus G3 looks like in April 2026. If it's a dud, then we can talk about the Command-Economy that's so seductive to FT contributors.
让我们看看2026年4月发布的擎天柱G3会是什么样子。如果它搞砸了，那我们就能聊聊这个让FT撰稿人如此着迷的计划经济模式了。
But if it's a hit - well maybe there's life in the Free-Market yet...
但如果它成功了——嗯，也许自由市场还有点生命力呢……
@Peasant vs Hillbilly
While Elon's strategy is Made in China!
马斯克的战略可是“中国制造”啊！
@But the truth
"A staff member, who asked not to be named, explained that these robots were not being used for collecting training data. 'These are the ones that perform when officials come to visit.'"
“一位不愿具名的员工解释说，这些机器人并非用于采集训练数据。‘这些是供领导参观时进行演示用的。’”
So great. All the advantages and disadvantages of Chinese culture summed up in a single paragraph.
太棒了。中国文化的所有优点和缺点，都浓缩在这短短一段话里了。
@Contrarian Investor
Yes, and they come to visit a lot.
是的，而且领导们经常来视察。
@Little Briton
I wonder why the researchers said “it remained unclear if it would yield the desired technological gains”? It seems reasonable that showing a robot exactly how to do things would enable it to learn how to deal with specific tasks. I suppose that enabling a robot to move about an unfamiliar environment, e.g., a house where it is expected to undertake cleaning and other tasks (but where the position of things may change from day to day), may require a much higher level of learning and capability.
我很好奇为什么研究人员会说“目前尚不清楚这是否能带来预期的技术收益”？向机器人精确展示如何操作，理应能让它学会处理特定任务，这听起来很合理。我猜想，要让机器人在不熟悉的环境中行动——比如一个需要它完成清洁等各项任务的住宅（而屋内物品的位置可能每天都在变动）——可能需要更高层次的学习能力和智能。
In any case it appears we are on the way to producing humanoid robots that will do useful work in homes, shops, offices, hospitality and factories. I suspect the genuinely useful home robot is at least 15 years in the future, but when it arrives it will be the most important consumer device in history. We must ensure that UK businesses and researchers are involved in the development.
无论如何，我们似乎正在朝着制造人形机器人的方向迈进，它们将在家庭、商店、办公室、服务场所以及工厂中承担有用的工作。我估计真正实用的家用机器人至少还要十五年才能问世，但当它到来时，将会成为历史上最重要的消费设备。我们必须确保英国的企业和研究人员能参与到这项开发中来。
@Brookfield One
$50,000 for a robot could be economically viable depending on the work that's done and where it's done, even assuming costs don't go down further in the future. If these robots can be trained to do office work or manufacturing while being able to fix itself, human workers will be in very serious trouble.
一台五万美元的机器人，根据工作内容和地点来看，在经济上可能是可行的——即便假设未来成本不再进一步下降。如果这些机器人能被训练去做办公室工作或制造业，同时还能自我维修，那人类劳动者可就真要大难临头了。
@Chenicus531
Perfect if your population is declining.
可如果你的人口正在减少，那就完美了。
@Digital-nomad
Have they consulted "The market"? Oh, they don't need to. I see. And that's why they can plan ahead. Got ya.
他们征求过“市场”的意见吗？哦，他们不需要。我懂了。所以他们才能提前规划。我明白了。
@Little Briton
Don’t worry, there will be a market. I expect that most of us would be prepared to pay upwards of £50,000 for a truly competent home robot. It would be a lot more useful than a car for many people.
别担心，总会有市场的。要我说，一台真正能干的家用机器人，咱们多数人估计都愿意掏五万英镑以上。对很多人来说，这玩意儿可比汽车实用多了。