Entrepreneurial Thought Leaders (ETL) - 史蒂夫·库辛斯(斯坦福大学)——现实世界中的机器人技术 封面

史蒂夫·库辛斯(斯坦福大学)——现实世界中的机器人技术

Steve Cousins (Stanford University) - Real-World Robotics

本集简介

史蒂夫·库森斯是斯坦福机器人中心的执行主任,也是Relay Robotics(前身为Savioke)的创始人,该公司为酒店和医疗行业研发自主服务机器人。库森斯的职业生涯体现了他致力于将机器人技术从研究推向实际应用的承诺。在本次演讲中,他分享了自己在研究、初创企业和学术界的经历,以提供关于招聘、机器人产品开发以及将机器人技术引入新行业的实用建议。 企业家思想领袖及其他斯坦福eCorner内容由斯坦福技术创业计划(STVP)制作,该计划是斯坦福工程学院的创业中心。STVP致力于赋能有志创业者,使其成为创造并规模化负责任创新的全球公民。 关注我们 YouTube:https://www.youtube.com/user/ecorner X:https://x.com/ECorner LinkedIn:https://www.linkedin.com/company/stanfordtechnologyventuresprogram/ Bluesky:https://bsky.app/profile/stanfordstvp.bsky.social 了解更多 STVP:https://stvp.stanford.edu/ eCorner由STVP提供:https://stvp.stanford.edu/ecorner 支持我们的使命:为全球学生和教育工作者免费提供斯坦福大学企业家思想领袖网络资源:https://stvp.stanford.edu/giving-to-stvp/

双语字幕

仅展示文本字幕,不包含中文音频;想边听边看,请使用 Bayt 播客 App。

Speaker 0

欢迎各位来到斯坦福大学的创业思想领袖研讨会。

Welcome everybody to the entrepreneurial thought leader seminar at Stanford University.

Speaker 0

创业思想领袖研讨会由BASIS(斯坦福创业学生商业协会)和STVP(斯坦福工程学院创业中心)联合主办。

The entrepreneurial thought leader seminar is brought to you by BASIS, the Business Association of Stanford entrepreneurial students, and STVP, Stanford Engineering's Entrepreneurship Center.

Speaker 0

我是拉维·维拉尼,管理科学与工程系的讲师,同时也是Alchemist企业初创加速器的负责人。

I'm Ravi Vilani, a lecturer in the management science and engineering department and the director of Alchemist and accelerator for enterprise startups.

Speaker 0

今天,我很荣幸地欢迎我们的下一位ETL主旨演讲嘉宾——史蒂夫·科因斯,作为百年庆典的一部分。

Today, I have the pleasure of welcoming our next ETL keynote as part of the centennial, Steve Cousins.

Speaker 0

史蒂夫延续了本学期我们庆祝工程学院百年华诞的传统,突出展示斯坦福工程学院在创业领域的卓越贡献。

Steve is continuing the tradition we have this quarter of celebrating the School of Engineering centennial by highlighting Stanford Engineering amazing contributions to entrepreneurship.

Speaker 0

史蒂夫对现代机器人生态系统的塑造,可能比大多数人意识到的还要深远。

And Steve has shaped the modern robotics ecosystem more than most people probably realize.

Speaker 0

史蒂夫是斯坦福机器人中心的执行主任,也是Relay Robotics的创始人,该公司原名Savvy Oak,致力于为酒店和医疗行业开发自主服务机器人。

Steve is the executive director of the Stanford Robotics Center and the founder of Relay Robotics, which was formerly Savvy Oak, which builds autonomous service robots for hotels and health care.

Speaker 0

史蒂夫职业生涯的主线始终是致力于推动机器人技术从研究走向实际应用。

The through line in Steve's career is he's always been passionate about pushing robotics from research to real world applications.

Speaker 0

他之前是Willow Garage的首席执行官兼总裁,该公司通过创建机器人操作系统(ROS)推动了开源机器人运动的发展。

He was previously the CEO and president of Willow Garage, which helped launch the open source robotics movement through the creation of ROS, the robot operating system.

Speaker 0

有多少人听说过ROS?

How many people have heard of ROS?

Speaker 0

好的。

Okay.

Speaker 0

这难道不是很棒吗?

So even if isn't that fantastic?

Speaker 0

太好了。

That's excellent.

Speaker 0

如果你没听说过它,那你很可能用过基于ROS构建的产品。

And if you haven't heard of it, you've probably used products that are built where ROS is really foundational.

Speaker 0

亚马逊,想想他们的仓储机器人。

Amazon, think about their warehouse robotics.

Speaker 0

波士顿动力、Cruise、NASA,它们的许多机器人都是基于ROS的技术开发的。

Boston Dynamics, Cruise, NASA, a lot of their robots are built on things that are coming out of Ross.

Speaker 0

史蒂夫还是开源机器人基金会的创始董事会成员,并于2017年荣获IEEE机器人与自动化发明与创业奖。

Steve's also the founding board member of the Open Source Robotics Foundation, and he's received the IEEE Award for Invention and Entrepreneurship in Robotics and Automation in 2017.

Speaker 0

他是一位三重计算机科学毕业生。

He is a triple computer science grad.

Speaker 0

有多少人是计算机科学专业的,或者正在考虑学习计算机科学?

How many people are CS majors or thinking about computer science?

Speaker 0

是的。

Yeah.

Speaker 0

好的。

Okay.

Speaker 0

所以,史蒂夫在华盛顿大学获得了计算机科学的学士和硕士学位,并在斯坦福大学获得了博士学位。

So Steve has his bachelor's and his master's in computer science from WashU and his PhD here from the farm, Stanford University.

Speaker 0

那么,不耽误时间了,有请史蒂夫·库辛斯来到ETL。

So without further ado, please welcome Steve Cousins to ETL.

Speaker 0

谢谢。

Thank you.

Speaker 1

你能听清楚我讲话吗?

Can you hear me okay?

Speaker 1

是的。

Yeah.

Speaker 0

你没问题。

You're good.

Speaker 1

好的。

Alright.

Speaker 1

谢谢拉维,能来到这里我感到非常荣幸。

Thank you, Ravi, and it's really a pleasure to be here.

Speaker 1

我看到关于这场演讲的公告时,就觉得这一定是一个了不起的团队,因此我很高兴能在这里开始。

I saw the announcement about this talk and I said, this has got to be an amazing group so I'm very happy to start here.

Speaker 1

我想先跟大家分享一系列故事,讲讲我是如何成为企业家的,或者我是如何走到今天的。

I wanted to start and I'm kind of going to tell you a series of stories that how I got to be an entrepreneur entrepreneur or how I kind of got to where I am.

Speaker 1

我会按时间顺序来讲,因为这样比较简单。

And I'm just going to go in chronological order because that seems simple.

Speaker 1

我通常做的演讲更侧重于机器人领域,这次我会尽量更聚焦于创业主题。

Normally I give talks that are more robotics focused, this one I'll try to be more entrepreneurial focused.

Speaker 1

我的职业生涯始于圣路易斯的华盛顿大学。

My career started at Washington U in St.

Speaker 1

圣路易斯。

Louis.

Speaker 1

我想说的是,我曾试图仅凭计算机科学硕士学位在大学里拥有一段不错的事业。

And the first thing I'll say is, I tried to have a nice career in a university with just a master's computer science.

Speaker 1

结果发现,如果你只有硕士学位,又想留在大学里,千万别这么做。

And it turns out, if you have a master's and you want to stay in a university, don't do it.

Speaker 1

大学里的职业发展路径几乎全是为博士设计的,因此大约四年之后,我就达到了自己能取得的最高点。

Basically, the career path in universities are all tuned for PhDs and so I had quickly, after about four years, hit the peak of what I was going to be able to do.

Speaker 1

我的老板告诉我:你应该考虑去读个博士。于是我决定,好吧,我去申请斯坦福和其他几所顶尖学校,如果能被其中一所录取,我就去;否则我就去找一份真正的工作。

And my boss told me, look, you should think about getting a PhD and I decided, okay, I'm going to apply to Stanford and a couple other top schools and if I get into one of them, I'll go and otherwise I'll get a real job.

Speaker 1

学术界确实是个很有趣的地方,那也是我想要的生活方式,但我同时也希望拥有一份事业。

Academia is kind of like a fun place to be and it was the lifestyle that I wanted, but I also wanted to have a career.

Speaker 1

我来跟你们讲一个发生在圣路易斯的故事。

One story I'll tell you from St.

Speaker 1

我在那里的医学院从事医学信息学工作了大约四年,当时我们雇了一名实习生。

Louis is I worked at the medical school there for about four years in medical informatics, and we hired this intern.

Speaker 1

所以第一个主题是关于招聘。

So the first theme will be about hiring.

Speaker 1

我们为那个夏天雇了这名实习生,这是一个由国家科学基金会资助的项目,报酬很低,每小时只有5美元。

We hired this intern for the summer, and it was a NSF funded program, so it paid, like, not very well, like $5 an hour for the summer.

Speaker 1

即使在当时,每小时5美元也不算多。

And even back then, 5 an hour wasn't a lot.

Speaker 1

他整个夏天都工作得很出色,非常非常优秀。

And so he worked for the summer and he was really, really, really good.

Speaker 1

优秀到我们问他:你知道吗?感恩节期间能回来工作一周吗?

Like so good that we said, You know what, can you come back at Thanksgiving just for the week?

Speaker 1

我们愿意每小时付你20美元。

We'll pay you $20 an hour.

Speaker 1

他回来了,说:太好了。

And he came back, he was like, Great.

Speaker 1

然后我们让他在圣诞节期间回来三周,时薪20美元,因为我们觉得他实在太优秀了,付给他的报酬比普通助研岗位还高。

And then we had him back for Christmas for all three weeks, dollars $20 an hour, and we just paid him more than he was going to get through normal assistantships because he was just that good.

Speaker 1

我讲这个故事,是因为后来在我的经历中,他又回来了,而那时我却为他工作了。

I tell you this story because he comes back later in my story where I worked for him.

Speaker 1

所以,这种事情真的发生了,对吧?

So this happens, right?

Speaker 1

你雇佣了某人,后来却可能反过来为他工作。

You work for somebody, you hire somebody, and then later you may end up working for them.

Speaker 1

像他这样的实习生,后来去了华盛顿大学,而那时我正好从圣路易斯来到斯坦福。

So keeping an intern like that, he ended up going to work at Wash U just at the time that I came out here to Stanford.

Speaker 1

在斯坦福,我作为一名计算机科学的研究生,正好有一个新项目启动,而获批的资助项目中包含了两个程序员或工程师的职位,用于协助研究工作。

And so here at Stanford, was working as graduate student in computer science and there was a project that was getting started and one of the things that the grant that got funded had a slot for two programmers, two engineers to help out with the research.

Speaker 1

就在那时,我已经在这里待了一年,可能是一年半,快两年了,当资助获批时,斯科特决定:好吧,他受够了圣路易斯。

And just about that time, I had been here for a year, maybe a year and a half, two years when the grant got funded, and Scott decided, okay, he had enough of St.

Speaker 1

路易斯搬到了加利福尼亚,他说:‘我能先住你这儿一段时间,想想接下来要做什么吗?’

Louis, he moved out here to California and he said, can I live with you for a little while and figure out what I'm going to do?

Speaker 1

我要去创业。

I'm going to do startup.

Speaker 1

我劝了他,我不知道这算好还是坏,我劝他不要去创业。

And I convinced him, I don't know if this was good or bad, I convinced him not to do a startup.

Speaker 1

我说:‘你该来斯坦福和我一起工作,因为我们之前在圣塔芭芭拉合作得特别开心。愉快。’

I said, You should come to Stanford and work with me because we had great fun working in St.

Speaker 1

路易斯。

Louis.

Speaker 1

来和我一起工作吧,你会遇到很多有趣的人,之后再创业也不迟。

Come work with me, you'll meet really interesting people and then you can do a startup.

Speaker 1

结果证明,这真是个非常好的建议。

And it turned out that that was actually really good advice.

Speaker 1

我的同学是拉里·佩奇,我知道谢尔盖稍后会在系列中发言。

My classmate was Larry Page, and I know Sergei is going to talk later in the series.

Speaker 1

那时我刚好在这里,而斯科特因为作为程序员来到斯坦福,得以帮助拉里和谢尔盖编写了谷歌的第一个版本——第一个网络爬虫。

This was the time that I just happened to be here, and Scott, because he came to Stanford as a programmer, got to help Larry and Sergei write the first version of Google, the first crawler.

Speaker 1

所以那时是 google.stanford。

So this was google.stanford.

Speaker 1

但他从未为谷歌工作过。

He never worked for Google.

Speaker 1

但这对他来说是一次绝佳的机会。

But it was a great opportunity for him.

Speaker 1

他是个非常非常有创业精神的人,所以当我毕业并进入人生下一阶段时,他已经开始了他的第一家初创公司的衍生项目。

And he was a very, very entrepreneurial guy, so by the time I graduated and went on to the next phase of my life, he had already was doing a spin off in his first startup.

Speaker 1

因此,他在我年轻时就成为了我眼中的创业者榜样。

So he was an example to me of an entrepreneur at a young age.

Speaker 1

那么,为什么我没有直接跳出去也创办一家初创公司呢?

Now, why didn't I just jump off and do a start up as well?

Speaker 1

我正在完成我的博士学位。

I'm finishing my PhD.

Speaker 1

我来斯坦福的时候有一个孩子,等我带着博士学位离开时,已经有了三个孩子。

It turns out when I came out to Stanford, I had one kid, and when I left Stanford with a PhD, had three kids.

Speaker 1

这并不理想。

So that's not ideal.

Speaker 1

这是一种有意识的生活选择。

This is sort of a conscious life choice.

Speaker 1

这不是偶然,我是有意为之的。

It wasn't an accident, I mean it was on purpose.

Speaker 1

但这确实是一个有意识的生活选择。

But this was a conscious life choice.

Speaker 1

当你有了孩子,孩子的幼儿园玩伴的父母就会成为你新朋友圈的一部分,这就是现实。

When you have kids, you start to meet, your kids have friends in preschool and you meet their parents and this becomes your new friends group, is the way it works.

Speaker 1

就像你上大学时会结识自己的朋友圈一样。

Just like when you go to college, you meet your friends group.

Speaker 1

于是我结识了其他几位父亲,发现很多做初创公司的人根本没时间陪孩子。

And so I was meeting these other dads and realizing that a lot of them who were in startups didn't see their kids very much.

Speaker 1

我不希望那样做,对吧?

And I didn't want to do that, right?

Speaker 1

所以我说的不是这好还是不好,只是这不是我想做的事。

So this was I'm not saying it's bad or good, it's just this is not what I wanted to do.

Speaker 1

所以我决定毕业后走工业研究这条路。

And so I decided when I graduated I'm going to go the industrial research route.

Speaker 1

于是我去了施乐帕洛阿尔托研究中心,那是个很有名的地方,你可能听说过,它是个人电脑的发源地等等。

So I went to Xerox PARC, which was a storied place that you may have heard of, beginning of the PC and so forth.

Speaker 1

于是,我通过斯科特间接体验了创业,看着他经营自己的初创公司。

And so, have sort of vicarious entrepreneurship through Scott, watching him do his startup company.

Speaker 1

与此同时,我走了另一条路,这么说吧。

Meanwhile, I went a different route, let's say.

Speaker 1

我在施乐帕洛阿尔托研究中心的选择是,本来可以去学术界工作,但again,我有三个孩子,而刚起步的学术职位收入似乎并不高。

Now, my choice at Xerox PARC was, could have gone to an academic job, but again, I had three kids and academic jobs in the beginning didn't seem to pay very well.

Speaker 1

我的另一个选择是Interval Research。

My other choice was Interval Research.

Speaker 1

我提到Interval,是因为Interval是由亿万富翁保罗·艾伦创立的一个有趣的地方。

And I mention Interval because Interval was this interesting place founded by Paul Allen, who's a billionaire.

Speaker 1

你可能知道他是微软的创始人之一。

You may know Bill Gates' partner in Microsoft.

Speaker 1

他创办了这家公司Interval,给了他十年的资金,说:去吧,像创业公司一样做这件事,但你是一个研究实验室。

And he founded this company, Interval, gave him money for ten years and said, Go off and do this like a startup, but you're a research lab.

Speaker 1

所以就像Xerox PARC,但没有Xerox。

So be Xerox PARC, but without the Xerox.

Speaker 1

去发明一些新东西,看看你能否把它做成一门生意。

And go off and invent some new things and see if you can make a go of it as a business.

Speaker 1

答案是否定的。

The answer was no.

Speaker 1

八年后,他们关闭了它。

After eight years, they shut it down.

Speaker 1

这给我上了一课。

That was a lesson for me.

Speaker 1

这些由亿万富翁资助的公司可能会突然关闭,而Interval就是这样一个例子。

Was like these billionaire funded companies might end abruptly, and that one happened.

Speaker 1

所以,当我还在Park时,我还看到了这张照片,这里你可以看到移动机器人和模块化机器人。

And so, while I was at Park, I also, this is a picture, here you can see the mobile robots, modular robots there.

Speaker 1

另一个实验室有个叫马克·伊姆的人,我有机会从研究转向高级开发,这属于这个主题的一部分。

There was a guy in another lab named Mark Yim and I had the chance to move from research to advanced development, which is part of this theme.

Speaker 1

我当时还没有进入机器人领域。

I wasn't yet in robotics.

Speaker 1

这可能是我第一次接触到有可能商业化的机器人技术。

This was probably my first brush with potentially commercial robotics.

Speaker 1

马克有一个想法,觉得他正在研发的模块化机器人可以做成一个有趣的玩具。

And Mark I'm had this idea that maybe this modular robots thing he was building would make a cool toy.

Speaker 1

所以我们试试看能不能把它商业化为一款玩具。

So let's see if we can commercialize it as a toy.

Speaker 1

我当时在高级开发部门,所以帮了他一把。

And I was in advanced development, so I'm helping him with that.

Speaker 1

当你试图将某物商业化为玩具时,你会去和公司谈。

When you're trying to commercialize something as a toy, you go talk to companies.

Speaker 1

玩具公司对事物有着非常独特的看法。

And the toy companies have a very interesting perspective on things.

Speaker 1

他们会直接告诉你,市场是如何结构的。

They would just tell you upfront, this is the way the market is structured.

Speaker 1

如果你把玩具授权给我们,你将获得大约5%或更少的分成。

If you license a toy to us, you're going to get, say, 5% or less.

Speaker 1

如果你基本上是把整个发明带给我们,你能指望的最高分成就是收入的5%。

The most you can hope for is 5% of the revenues revenues if you're basically bringing us the whole invention.

Speaker 1

我们当时说:真的吗?

We're like, really?

Speaker 1

是的。

Yeah.

Speaker 1

制造玩具的成本占25%。

The cost of making the toy is 25%.

Speaker 1

制造该产品的公司会获得接下来的25%。

The company that's making it gets the next 25%.

Speaker 1

所以,如果你把材料成本乘以四,那就是你在商店里需要支付的目标价格。

So if you take the cost of bill materials and you multiply it by four, that's going to be the target price point that you would pay in a store.

Speaker 1

了解这一行业结构的教训非常重要。

Learning that lesson was really, really important about the structure of the industry.

Speaker 1

我们发现,如果我们想制造这些模块化机器人,并且想达到生日礼物的价格——这是90年代时人们去参加生日派对时愿意为玩具支付的典型价格——我们就必须只配一个模块。

Figure out that if we want to make these modular robots and we want to hit the birthday price this was in the '90s birthday price meaning the price that people would pay for a birthday present if you're going to a birthday party, which is like the target most price that you would pay for toys, we had to basically send it with one module.

Speaker 1

一个模块。

One module.

Speaker 1

现在,一个模块的模块化机器人系统相当无聊,他们进行了用户测试。

Now, one module of modular robot system is pretty boring, and they play tested it.

Speaker 1

他们制作了广告,并将其与遥控汽车进行对比。

They created ads and they showed them against radio controlled cars.

Speaker 1

哇,看看这些车跳来跳去,太酷了。

Wow, look at these cars jumping over and all this stuff.

Speaker 1

然后这就是这个模块化机器人。

And then here's this modular robot.

Speaker 1

我们尽力把它做得酷一些。

We tried to make it as cool as we could.

Speaker 1

孩子们不买账,玩具公司也不感兴趣,这个项目就没了。

The kids didn't bite, the toy companies didn't bite, that project went away.

Speaker 1

这让我学到一个关于如何将产品推向市场的重要教训。

An important lesson for me about how you get things to market.

Speaker 1

我在IBM研究院待了一段时间,我提到这一点是因为IBM非常注重流程。

I spent a little bit of time at IBM Research, and the reason I mentioned this is that IBM is very big on process.

Speaker 1

他们对每件事都有流程。

They have process for everything.

Speaker 1

他们其中一个流程就是你的职业发展流程。

And one of the processes they have is your career process.

Speaker 1

所以有一天,我的老板丹·希夫曼来找我,说:我得跟你开个会。

And so at some point, my boss, Dan Shiffman, came to me and he said, I have to have this meeting with you.

Speaker 1

我们得谈谈你的职业发展。

We have to talk about your career.

Speaker 1

你希望十年后在哪里?

Where do you want to be in ten years?

Speaker 1

我得记录下我们有过这次对话,而且你需要签字。

I have to write down that we had this conversation and you have to sign it.

Speaker 1

好吧,这就是流程。

Okay, that's the process.

Speaker 1

但有趣的是我对他说的话。

But what's interesting is what I said to him.

Speaker 1

我说:嗯,十年后,也许十一年后,当我的小儿子上大学时,我想成为一名创业者,或许当一家公司的首席执行官。

I said, Well, in ten years, maybe in eleven years, when my youngest is going to go to college, that's when I want to be an entrepreneur and maybe be a CEO of a company.

Speaker 1

他跟我说了一些非常有意思的话。

And he said something really interesting to me.

Speaker 1

他说:我也曾想过当首席执行官。

He said, I thought about being a CEO too.

Speaker 1

当我认真思考了很久之后,我意识到自己并不具备胜任这份工作所需的道德弹性。

And when I really thought about it a lot, I realized that I don't have the moral flexibility required for that job.

Speaker 1

我当时就想:什么?

And I was like, What?

Speaker 1

道德弹性?

Moral flexibility?

Speaker 1

尽管如此,我后来还是接受了首席执行官的职位,但这个想法确实让我有了一个有趣而深刻的领悟。

I went ahead and took CEO jobs later anyway, but it was sort of an interesting, eye opening way to think.

Speaker 1

好了,以上都是关于Willow Garage的铺垫。

All right, so that's all kind of preamble to Willow Garage.

Speaker 1

Willow Garage是个非常特别的地方,不是一家典型的初创公司。

Willow Garage was a really interesting place, not a typical startup.

Speaker 1

我一开始讲的是Scott Hasson的故事,他是我雇来的实习生,后来我把他带到了这里。

I started off the story with Scott Hasson, who's the friend that I hired as an intern and then moved out here.

Speaker 1

他走的是另一条路,创办了一家成功的初创公司eGroups,后来被雅虎收购,变成了Yahoo Groups,因为他认识拉里和谢尔盖,所以得以早期投资谷歌,拿到了创始股份。

Well, took his path, had a successful startup company called eGroups, which Yahoo bought, became Yahoo Groups, and then he was able to invest in Google early because he knew Larry and Sergey, and so he got those founder shares.

Speaker 1

于是,斯科特突然有了很多钱,他回到帕洛阿尔托,说:‘我买了一栋楼,想创办一家公司,你愿意加入吗?’

And so suddenly Scott had a lot of money, and he came back to Palo Alto and he said, I bought a building, and I want to start a company, and do you want to join?

Speaker 1

他带我看了那栋楼,感觉还行。

And he showed me the building, and it was okay.

Speaker 1

但我当时说:‘不。’

But I was like, no.

Speaker 1

他问:‘为什么不行?’

And he goes, Why not?

Speaker 1

我说:‘你都没告诉我你想做什么,我怎么知道能不能帮上忙?’

And I said, Well, you haven't even told me what you want to do, so how do I know if I can help you?

Speaker 1

这似乎让他很惊讶。

And that seemed to surprise him.

Speaker 1

于是他去想了几个星期,回来后说:‘好吧,我想好了。’

And so he went off and thought about it for a couple of weeks, and he came back and he said, Okay, I thought about it.

Speaker 1

自动驾驶汽车真的很酷。

Self driving cars are really cool.

Speaker 1

那是2006年。

This was 2006.

Speaker 1

自动驾驶汽车真的很酷。

Self driving cars are really cool.

Speaker 1

我们刚刚经历了DARPA大挑战赛,你们都知道斯坦福在2005年赢了。

We just had the DARPA Grand Challenge that you all know Stanford won in 2005.

Speaker 1

DARPA城市挑战赛将在2007年举行。

The DARPA Urban Challenge was going to happen in 2007.

Speaker 1

他对此非常兴奋,说服了我:有些团队由大学赞助,有些是独立团队,而他干脆买下了一个独立团队。

He got so excited about this that he convinced there were teams that were sponsored by universities and then there were some teams that were independent, and he basically bought an independent team.

Speaker 1

换句话说,他招聘了人员,买了三辆车,把他们安置在自己正在建造的那栋楼里,启动了自动驾驶汽车项目。

In other words, he went, he hired the people, and he bought them three cars, and he put them into this building that he was building and said, self driving car project.

Speaker 1

然后他问我:你想加入吗?

And he said, do you want to join that?

Speaker 1

我说:好吧,我可以做这个。

And I was like, okay, I can do that.

Speaker 1

他问我:‘你想要什么职位?’

And he said, what job do you want?

Speaker 1

我说:‘首席执行官。’

And I said, CEO.

Speaker 1

他回答:‘好的。’

And he goes, okay.

Speaker 1

那是我的第一次首席执行官面试。

That was my first CEO interview.

Speaker 1

这和我原本预期的方式并不一样。

It's not exactly how I expected it would happen.

Speaker 1

但Willow Garage很有趣。

But Willow Garage was interesting.

Speaker 1

它既不是研究实验室,也不是营利性公司。

It was not a research lab and it was not a for profit company.

Speaker 1

我们有一个信条:先追求影响力。

We had this mantra, let's do impact first.

Speaker 1

我们先关注影响力,再考虑资本回报。

We'll worry about impact first and we'll worry about return on capital second.

Speaker 1

我们有一个想法,就是孵化公司,让这些公司最终积累价值,这样Willow Garage就能保留它们的一部分股份,从而为我们的所有工作创造收益。

And we had this idea that we would spin off companies and the companies would eventually accrete value and that's how Will Garage would keep some share in them, and that's how we would basically make money for all this work.

Speaker 1

如果这不是一个研究实验室,也不是试图开发产品的公司,那这个地方到底是什么?

If it's not a research lab and it's not trying to make a product, what was this place?

Speaker 1

最后,我们决定最好把它称为一个智库,但这只是个简称。

In the end, we decided that the best I sometimes called it a think tank, but that's just a shorthand.

Speaker 1

它其实并不是一个智库。

It's not really a think tank.

Speaker 1

我们称它为迷途博士们在创业途中暂居的中转站,这可能是最准确的描述。

We called it a halfway house for wayward PhDs on their way to start which actually is probably the most accurate.

Speaker 1

许多初创公司都源于曾来到Willow Garage后离开的人。

There's a lot of startups that came of people who came to Willow Garage and moved on.

Speaker 1

我们制造了这台机器人,你看到它在那里打台球了。

We built this robot that you see playing pool there.

Speaker 1

玩台球这个项目只花了一周时间。

The pool playing was a one week project.

Speaker 1

有了机器人和软件之后,这只是一个证明你能够做到的演示。

After you had the robot and the software, it just was a proof that you could do it.

Speaker 1

还有一个项目解决了社交机器人或家用机器人的根本问题,那就是‘给我拿瓶啤酒来’。

There was another project which addressed the fundamental problem of social robots or robots in the home, which is bring me a beer.

Speaker 1

这也是一个为期一周的项目。

That was another one week project.

Speaker 1

而且这个机器人实际上相当复杂。

And the robot was actually pretty sophisticated.

Speaker 1

它能打开冰箱,取出啤酒,再拿过来。

Like, could open the refrigerator, get a beer, bring it.

Speaker 1

这发生在十五年前,那时还没有如今这类人形机器人。

And this was fifteen years ago before we had today's humanoids.

Speaker 1

我们制造了大量这样的机器人。

And we made a lot of them.

Speaker 1

我们把它们送给了大学。

We gave them away to universities.

Speaker 1

这是其中一件有趣的事。

This is one of the things that was interesting.

Speaker 1

斯科特坚持认为我们需要一个出色的实习生项目。

Scott was adamant that we wanted to have an amazing intern program.

Speaker 1

第一年的实习生团队,他带我们体验了零重力飞行,就是那种做抛物线飞行、真正体验零重力的航班。

And the first year intern class, he took us on zero g, which is this flight where you actually do the parabolas and you have zero g.

Speaker 1

实习生们都很喜欢,除了那些吐了的人。

And the interns loved it, except for the ones who threw up.

Speaker 1

那是一个有趣的时期,因为我们学到了另一个非常重要的教训。

And it was an interesting time because we learned another really important lesson.

Speaker 1

我们启动了自动驾驶汽车项目。

We had started the self driving car project.

Speaker 1

斯科特还有另一个项目,是一个自主航行的船。

Scott had another project which was an autonomous boat.

Speaker 1

然后我觉得我们需要丰富我们的产品线,于是我来到斯坦福,招募了一些正在研究斯坦福人工智能机器人的学生,他们过来参与了PR2机器人的开发,就是你们看到的那台机器人。

And then I thought we need to round out the portfolio, and I came over here to Stanford and recruited some students who were working on the Stanford AI robot, and they came over to build the PR2 robot, that robot that you saw.

Speaker 1

我们很快就开始讨论专注的问题,对吧?

Very quickly, we had a conversation about focus, right?

Speaker 1

因为我们当时正在同时推进三个项目。

Because we're trying to do three projects.

Speaker 1

我之前来自施乐帕洛阿尔托研究中心和IBM研究院,那里每个人通常负责两个项目,对吧?

Now, I came out of Xerox PARC and IBM Research where you had two projects per person, right?

Speaker 1

因为组织奖励的是启动项目,所以你会得到你所奖励的东西。

Because the organization's rewarded starting projects, and so that you get what you reward.

Speaker 1

但在这里,我们决定专注于一件事,并且要把这一件事做到极致。

But here, we said we want to focus on one thing, and we're going to do one thing well.

Speaker 1

我们试图招聘所有领导斯坦福自动驾驶汽车项目的人,但都被谷歌挖走了。

We tried to hire all the people who led the Stanford car project, project, but Google got them all.

Speaker 1

因此,我们决定专注于个人机器人。

And so instead, we decided to focus on the personal robot.

Speaker 1

关于机器人领域的发展,有趣的是,你建造了一台能力超强的机器人。

The interesting thing about, this was kind of getting going in robotics, you build a robot that is super capable.

Speaker 1

我们的想法是,全球这些公司和研究实验室其实都在重复造轮子,因为它们都得自己开发硬件,然后再为硬件开发软件。

Our idea was all these companies, all these research labs around the world were basically reinventing the wheel, because they all had to basically build their own hardware and then build their own software for it.

Speaker 1

这个项目的理念是,我们向众多大学提供统一的硬件,让它们都能在同一项目上合作。

And the idea of this project was let's give out common hardware to a lot of different universities so that they can all work on the same project.

Speaker 1

然后我们还有一个很棒的实习生项目,比如第一个夏天,我们有15名实习生。

And then we had this amazing intern program where we would have, like the first summer, we had 15 interns.

Speaker 1

而我们当时只有17名员工。

We only had 17 employees.

Speaker 1

这个比例相当惊人。

So it's a pretty amazing ratio.

Speaker 1

实习生项目最棒的地方在于,它最终运作得恰好

So the amazing thing about the intern program was it ended up working exactly

Speaker 0

像疫情期间一样。

like the pandemic.

Speaker 0

我们会从最好的实验室带来博士生,因为他们都想要这些机器人。

We would bring PhD students from the best labs because they all wanted the robots.

Speaker 0

所以第一年,当我们承诺制造机器人时,世界各地的顶尖教授都派出了他们最优秀的学生来参与这个项目。

So the first year when we were promising to build robots, top professors from around the world were sending their best grad students to be part of this

Speaker 1

实习生项目。

intern intern program.

Speaker 1

之后,实习生们到来,一起合作开发这个开源的机器人操作系统,而这就是他们能使用的最佳平台。

And after that, the interns came, they all worked on this robot operating system, open source thing together, and then that's the best platform they could work on.

Speaker 1

于是他们把这套系统带回了自己的大学,并告诉其他研究生:嘿,这是开源的。

So they took it back with them to their universities, and then they told the other grad students, hey, this is open source.

Speaker 1

它运行得非常好。

It works really well.

Speaker 1

你们应该使用它。

You should use it.

Speaker 1

同时,我们也在招聘大量工程师来帮助开发它。

Meanwhile, we're hiring a lot of engineers to help build it.

Speaker 1

经过大约三年的这种反复传播,这个想法、这个萌芽,就像一场大流行病一样扩散开来,因为学生们毕业后会去其他大学,并告诉他们的朋友。

And so after about three years of this back and forth, that idea, that germ, that basically acted like a pandemic because the students would graduate, they would move on to other universities, they would tell their friends.

Speaker 1

所以很快,全世界都在使用这款开源软件。

And so very quickly, the whole world was using this open source software.

Speaker 1

这并不是说没有其他选择。

And it wasn't like there was no alternative.

Speaker 1

确实有一些替代方案,但它们都相当简单。

There were some alternatives, but they were pretty simple.

Speaker 1

比如,某所大学的三个研究生做了一个东西并开源了。

They were like three grad students at a university had made something and open sourced it.

Speaker 1

但如果你要和这样的项目竞争,而你有一支由15名全职工程师组成的团队,你的系统显然更加完善。

But if you're competing with that and you've got a team of 15 full time engineers, your system is much more well engineered.

Speaker 1

因此,随着开源生态的发展,那里取得了许多成功。

And so you can see as the open source world has grown on, lot there's of success there.

Speaker 1

所以,这个机器人操作系统是所有这些学生共同贡献,加上我们团队核心工程工作的成果,迅速成为了全球标准。

So the robot operating system was contributions from all these students together with the core engineering stuff that we built and quickly became the standard around the world.

Speaker 1

所以,当我想到Willow时,我们当时只有85人,公司只存在了六年,却有130名实习生,如果从这个角度看,每年的实习生数量相当可观。

So, when I think about Willow, we had only 85 people and it was only around for six years, and we had 130 interns, which is a lot of interns per year if you think of it that way.

Speaker 1

我们孵化了八家衍生公司,其中三家是非营利组织,三家是营利性企业。

We had, eight spin off companies, three of them were non profits, three of them were for profits.

Speaker 1

谷歌收购了其中大多数公司,你还记得2012年谷歌收购了一大批公司吗?

Google snapped up most of them in this Remember when Google bought up a bunch of companies in 2012?

Speaker 1

其中很多就是这些公司。

A lot of them were these companies.

Speaker 1

而PR2机器人本身,我们最终制造了50台,并将它们部署到全球各地的大学。

And, the PR2 itself, we ended up making 50 robots and putting them universities around the world.

Speaker 1

即使过了十五年,仍有一些在使用中,这在硬件领域其实相当了不起。

There are some that are still in use even after fifteen years, which is pretty amazing for hardware, actually.

Speaker 1

大多数仍在使用的机器人,都已经把原来的电脑和传感器换成了更现代的版本。

Most of the ones that are still in use, they've ripped out the computers and put in more modern computers and sensors.

Speaker 1

而机器人操作系统至今仍在持续发展。

And the robot operating system continues to go on.

Speaker 1

每年都会举办一次罗斯·康会议。

There's a Ross Conn conference that happens every year.

Speaker 1

在过去两年里,这个会议在新加坡吸引了将近1000人参加。

In the last two years, it's had almost 1,000 people, Singapore.

Speaker 1

所以有大量国际人士前来参与。

So people are coming internationally in big numbers.

Speaker 1

这是一个成功的项目,持续对机器人领域产生重大影响。

So this this is a successful project that continues to have big impact on robotics.

Speaker 1

但正如我所说,这是一家奇特的初创公司,威尔开始时,我们有资金。

But as I said, this was this weird startup where start Will started, there was We had money.

Speaker 1

我们没有名字。

We had no name.

Speaker 1

我们没有人员。

We had no people.

Speaker 1

对吧?

Right?

Speaker 1

为了启动Will Garage,我尝试招聘我的第一位工程师。

And so in order to get Will Garage going, I tried to hire my first engineer.

Speaker 1

我实际上联系了参与玩具项目的马克·耶姆,问他我该雇谁。

I actually called Mark Yim, who was on the toy project, and I said, who should I hire?

Speaker 1

他说:‘我有个在宾夕法尼亚大学的博士后,你应该雇他。’

And he said, oh, I've got this post doc at Penn, you should hire him.

Speaker 1

我们面试了他,觉得他很棒。

And we interviewed him and we said great.

Speaker 1

他也同意愿意加入我们。

And he agreed he wanted to come.

Speaker 1

但他后来打电话来说,他的导师们建议他不要来,因为你们根本没人听说过。

And then he called us and said, My advisors are telling me not to come because you're nobody, basically.

Speaker 1

他们没听过你们的名字,不想让我浪费自己的职业生涯,所以他决定去做别的事。

They never heard of you, they don't want me to throw my career away, so I'm gonna go do something else.

Speaker 1

所以,就像当初斯科特还是本科生时,我和凯伦·费德姆费了很大劲才把他招进实验室一样。

And so, like with Scott, when he was an undergrad, worked really hard to get him and Karen Fiedem to get that first person in that lab.

Speaker 1

我又做了同样的事情。

Again, I did the same thing.

Speaker 1

这次我飞到了费城,做了一场演讲,逐一与所有导师交谈,告诉他们:这可是正经事。

This time I flew to Philadelphia, I gave a talk, I talked to all the advisors one on one and I said, Hey, this is legit.

Speaker 1

我有斯坦福的博士学位,肯定够聪明。

Have a PhD from Stanford, I must be smart.

Speaker 1

而且我们有大把大把的钱,聪明人很重要,所以你应该推荐他来。

And we have lots and lots of money and money and smart is good, so you should send him.

Speaker 1

不知怎么的,他们同意了。

And somehow they said okay.

Speaker 1

然后我们就开始招聘了。

And then we started hiring.

Speaker 1

我们还办了一场招聘午餐会。

The other thing we did, we had a recruiting lunch.

Speaker 1

我们邀请了各种斯坦福的机器人专业学生。

We invited all kinds of Stanford robotics students.

Speaker 1

没人来,但一位兼职教员加里·布拉德斯基真的来了。

None of them came, but one of the adjunct faculty, Gary Bradsky, actually did come.

Speaker 1

所以那次午餐我们没招到学生,但却成功吸引了一位教员。

So we didn't succeed in grabbing a student at that first lunch, but we managed to grab a faculty instead.

Speaker 1

加里有很多人脉资源。

And then Gary had all kinds of connections.

Speaker 1

他就是开发OpenCV的那个人。

He was the guy who did Open CV.

Speaker 1

这让我们再次开始引进各种优秀人才。

That caused us to basically start, again, bringing in all kinds of great people.

Speaker 1

实习项目的另一点是,第一年后,由于我们招聘的都是高年级研究生,很多人即将毕业。

The other thing about the intern program was that after the first year since we had hired senior grad students, a lot of them were graduating.

Speaker 1

在这15名学生中,我们最终重新聘用了大约九人,要么是第二次实习,要么直接转为全职员工。

And we ended up, out of those 15 students, hiring back about nine of them, either for a second internship or a number of them as full time employees.

Speaker 1

到WIL结束时,仅仅六年之后,其中三四人已经进入了核心团队。

And by the end of WIL, after only six years, three or four of them were on the senior team.

Speaker 1

他们现在已经是WIL的资深人员了。

They were now the senior people at WIL.

Speaker 1

所以,这是一种非常有趣的方式来让事情启动起来。

So it was a really interesting way to kind of get something going.

Speaker 1

但正如我所说,这并不是一个真正的初创公司。

But as I said, it wasn't a real startup.

Speaker 1

我不需要去筹集任何资金。

I didn't have to raise any money.

Speaker 1

在我看来,一个真正的初创公司意味着你必须筹集资金,或者必须以某种方式赚钱。

To me, a real startup means you've to raise money, or you have to somehow make money.

Speaker 1

这家公司两者都没做。

This company was doing neither.

Speaker 1

于是我创办了Savioque,他问起Savioque这个名字的由来,因为大家都叫它Savioque。

So, I started Savioque and the story, he was asking about the name Savioque because everybody calls it Savioque.

Speaker 1

我们基本上给它取了这个名字,当时说:好吧,为了把这件事推起来,我们需要一些现金。

We basically named it, we said, okay, in order to get this thing off the ground, we need some cash.

Speaker 1

威洛实验室之前谈妥了一份合同,但斯科特说:别签那份合同,我现在就要关停这个项目。

There was a contract that Willow Garage had negotiated, but Scott said, Well, don't take that contract because I'm going to shut this thing down now.

Speaker 1

他决定专注于其中一家衍生公司,不再为威洛实验室提供资金。

He decided he wanted to focus on one of the spin offs and not fund Willow anymore.

Speaker 1

所以,这是第二家由亿万富翁资助却突然终止的公司。

A So, second billionaire funded company that ended abruptly.

Speaker 1

这正在形成一种模式。

So, this is becoming a pattern.

Speaker 1

于是斯科特说:我们要把这个项目停掉。

So, Scott was saying, We're gonna put this down.

Speaker 1

我问他:我能接手这个资助吗?

I said, Can I take that grant?

Speaker 1

他回答:是的,你可以拿走这个资助。

And said, Yeah, you can get that grant.

Speaker 1

于是,我联系了那家公司,他们说:当然可以,你去做吧。

So in order, I called the company, they said, Yeah, you can do it.

Speaker 1

我该怎么把这份合同转到我的新公司名下?

How do I get that contract assigned to my new company?

Speaker 1

首先,你得有个银行账户。

Well, you have to have a bank account.

Speaker 1

就这么简单。

That's all.

Speaker 1

好的,签这些文件,然后把你的银行账号给我们。

Okay, so sign these papers and give us your bank account number.

Speaker 1

我该怎么开一个银行账户?

How do I get a bank account?

Speaker 1

啊,我得先有个公司,那我该怎么成立一家公司呢?

Ah, I need a company, and how do I get a company?

Speaker 1

我需要一个名字。

I need a name.

Speaker 1

就这样了。

So that was it.

Speaker 1

现在我们需要确定一系列步骤。

Now we gotta get a sequence of things.

Speaker 1

你怎么取名字?

How do you get a name?

Speaker 1

一群人围在房间里,对着白板,一起想个名字。

A bunch of people in a room, whiteboard, let's make up a name.

Speaker 1

我们白板上写了一大堆AI相关的词,比如智能、聪明、机敏。

We had a whole bunch of AI words on the board like smart, clever, savvy.

Speaker 1

我们保留了‘机敏’这个词。

Savvy, we held onto.

Speaker 1

然后我们还有另一个栏目,写了一堆树的名字。

And then we had a bunch of tree names is another column.

Speaker 1

我们只是在头脑风暴。

We were just brainstorming.

Speaker 1

在某个时刻,我看着白板说:‘Savvy Oak(机敏橡树)’。

At And some point, I looked at the board and said Savvy Oak.

Speaker 1

就这样定了。

And that was it.

Speaker 1

我们故意拼得奇怪,谷歌上搜不到任何结果,看来是个好名字。

We spelled it funny, zero hits on Google, must be a good name.

Speaker 1

但我们没注意到,人们会把它念成 Savvy okey,而不是 karaoke。

Except we didn't catch that people would call it Savvy okey instead of karaoke.

Speaker 1

于是我们创办了这家公司,并有了这个想法。

So we started this company and we had this idea.

Speaker 1

实际上,我们起步时的情况和 Willow 不同——Willow 刚开始时几乎没人,但资金充裕;而我们只有很少一点资金,几个曾就职于 Willow 且想继续合作的优秀人才,还有一些模糊的想法,觉得可以在科技领域做点什么。

We actually started with, unlike Willow which started with essentially no people and a lot of money, here we had a little tiny bit of money, a few really good people who had been at Willow and who wanted to work together and some vague ideas that we could do something in technology.

Speaker 1

这正是别人告诉你的不该做的事。

So this is like what they tell you you shouldn't do.

Speaker 1

先有技术,再试图创办公司。

Start with a technology and try to make a company.

Speaker 1

但我们已经有了公司和团队,所以我们就说,好吧,我们试试看。

But we had a company, we had a team, so we said, Okay, we're going to try it.

Speaker 1

我们知道我们需要集中精力。

We knew that we needed to focus.

Speaker 1

我们知道需要找到产品与市场的契合点,于是我们从酒店业入手,实际上我们最初的想法是服务行业。

We knew that we needed to come up with a product market fit, and we started with the hotel We actually had an idea of the service industry.

Speaker 1

这其实是我的一个误解。

So this is actually a misconception on my part.

Speaker 1

或者说,我不一定有遗憾,但我会告诉你我现在明白了什么。

Or, I don't know, I don't necessarily have regrets, but I'll tell you what I now know.

Speaker 1

我以为物流很无聊。

I thought that logistics was boring.

Speaker 1

我只是觉得它很无聊,所以我说,我不想做移动机器人和物流。

I just thought it was boring, and so I said, Well, I don't want to build mobile robots and logistics.

Speaker 1

我想进入服务行业,进入一片蓝海。

I want to go into the service industry, into a greenfield.

Speaker 1

蓝海听起来很棒。

Greenfield sounds great.

Speaker 1

绿色领域确实很酷,但如果你想赚钱,就得解决如何让这个市场运转起来的问题。

Greenfield is really cool, except for if you want to make money, because you to solve the green you have to figure out how to make that market work.

Speaker 1

与此同时,物流领域卖出了大量的机器人,因此许多从事与我们类似技术的公司都取得了成功,因为它们能一次性卖出很多台机器人。

Meanwhile, logistics sold lots and lots of robots, and so lots of companies that were also doing very similar technology to what we did had success because they could sell many robots at a time.

Speaker 1

我们选择了另一条路,决定进军酒店行业。

We went a different route, and we said, let's go after hotels.

Speaker 1

我认为我们也能一次性卖出很多台机器人。

I think we'll be able to sell a lot of robots at a time.

Speaker 1

这是基于对酒店行业的误解。

This was based on a misunderstanding of the hotel industry.

Speaker 1

这个故事,以及我想传达给你们所有人的心得是:当你打算创业并进入一个行业时,你需要深入了解这个行业。

And the story that, and sort of the message for all of you, is when you're going to do a start up and you're going to go into an industry, you need to learn a lot about that industry.

Speaker 1

你需要找到一个在该行业工作的人,让他们告诉你这个行业究竟是如何运作的。

You need to find somebody who works in that industry who can tell you how it actually works.

Speaker 1

如果你在推广某个产品或想法,你可以去和人们交谈。

And you can go talk to people if you're pushing a product or an idea.

展开剩余字幕(还有 229 条)
Speaker 1

我们做了一个故事板,然后去和人们交流,他们对此都非常兴奋。

We had a storyboard, we went and talked to people, they got all excited about it.

Speaker 1

我们让一家大型酒店品牌的副总裁非常激动。

We got a vice president of one of the major hotel brands super excited.

Speaker 1

我现在理解他为什么兴奋了,但并不是因为我们能赚很多钱。

And I understand why he was excited now, but it wasn't because we were going to make a lot of money.

Speaker 1

他兴奋是因为他能获得大量公关宣传,因为机器人上镜效果很好。

He was excited because he was going to get a lot of PR, because robots are great on camera.

Speaker 1

他立刻就看明白了。

And he saw that immediately.

Speaker 1

我没看出这一点。

I didn't see that.

Speaker 1

我以为,天啊,我们要做点大事了。

I thought, woah, we're going to do something.

Speaker 1

我们要为你创造一些真正有用的东西。

We're going to create something really useful for you.

Speaker 1

所以我们实际上把这个机器人——上面那张照片是我和四季度酒店里的Relay机器人在一起的照片。

So we put this robot actually, picture up there is me with a relay robot in the Four Seasons Hotel here in Palo Alto.

Speaker 1

四季度酒店是我们真正让这个东西运行起来的第一个地方。

And Four Seasons was the first place that we actually made this thing work.

Speaker 1

他们允许我们进入他们的电梯间,摆弄他们的电梯,而事实证明,别人都不会让你这么做。

They let us go into their elevator room and muck around with their elevators, which it turns out nobody else will let you do.

Speaker 1

我会谈谈这一点。

I'll talk about that.

Speaker 1

所以我们让机器人实现了自主导航。

So we got the robot to navigate autonomously.

Speaker 1

这是重大的技术突破,我们能够实现周围环境的导航。

This was the big technical achievement, we could navigate around.

Speaker 1

我们与谷歌风投合作,他们有一个设计团队,帮助我们改进用户界面和用户体验。

We worked with Google Ventures, they had a design group that helped us work on the user interface, the user experience.

Speaker 1

所以,基本上我们有了整个系统,嗯,酒店房间里有一部电话。

So we had this whole thing basically, well, there's a telephone, in the hotel room.

Speaker 1

这个机器人的理念是,你拿起电话,拨打前台,说:嘿,能给我送点东西吗?

The idea of this robot is that you would pick up the phone, you would call the front desk and say, hey, would you send me something?

Speaker 1

我们最初的想法是,比如给我送一把牙刷。

Send me a We originally thought, like, send me a toothbrush.

Speaker 1

实际上没人想要牙刷,而且它也几乎不值钱。

Nobody actually wants a toothbrush and it's worth, like, nothing.

Speaker 1

结果发现,这根本不是一个靠谱的商业点子,这也是我们学到的另一件事。

So it turns out to be like a really bad pitch for a company, which was another thing that we learned.

Speaker 1

但机器人能把东西送到你面前,这一点确实很特别——曾经有个人,我们亲眼看着,因为一开始我们只是在测试这东西能不能正常运行,当时机器人正停在他房间门口。

But the fact that the robot would bring it to you, I mean, there was somebody who literally we watched because in the beginning we were testing whether the thing actually was going to work and it was at somebody's door.

Speaker 1

我们躲在角落里偷偷观察。

We were kind of hiding around the corner.

Speaker 1

我们听到门开了,看到我们设计的机器人体验,那人脱口而出:天哪,这太棒了。

We heard the door open and this robot experience that we had designed, and the guy said, holy shit, this is amazing.

Speaker 1

很多客人因此感到非常兴奋,因为他们以前从未见过机器人。

And a lot of excitement in guests because they had never seen a robot before.

Speaker 1

这是一种新奇效应。

This is a novelty effect.

Speaker 1

这并不能成为一项好生意的基础。

This is not the basis of a good business.

Speaker 1

所以这一切中最大的障碍是电梯,因此我把这个演讲命名为‘电梯故事’。

So had the biggest hurdle in all of this was elevators, and so that's why I called the talk elevator stories.

Speaker 1

在酒店里,人们第一个问的问题是:我们该如何真正改造电梯?

The first question that people asked in hotels, we said, well, how do we actually change the elevators?

Speaker 1

我提到在四季酒店,他们原本不会让我们改造,但实际上他们确实允许我们修改电梯。

And I mentioned in Four Seasons that they weren't going to let us, that they did actually let us modify the elevators.

Speaker 1

但其他酒店都不会这么做。

But no other hotels were going to do that.

Speaker 1

原因是,电梯被视为关键安全系统。

And the reason is, elevators are considered a safety critical system.

Speaker 1

你不会允许一家机器人公司进来试用他们的产品,去改动你的电梯,从而可能影响你的客人。

You're not going to let some robot company that's coming in to try out their product mess with your elevators, and potentially interfere with your guests.

Speaker 1

我不明白为什么四季酒店允许我们这么做。

I don't know why the Four Seasons let us do that.

Speaker 1

但他们确实允许了。

But they did.

Speaker 1

我们接下来合作的另一个品牌,来自东海岸的大品牌,他们说你们可以做,但必须通过电梯公司来操作。

The next brand that we were working with, the big brand from the East Coast, they were saying, you you can do it, but you've got to work through the elevator company.

Speaker 1

于是我们开始了解一些关于市场结构的知识,这是一堂非常重要的课。

So we started to learn some things about structures of markets that was a pretty important lesson.

Speaker 1

首先是酒店,我们原本以为会卖出很多机器人,因为我们打算去万豪,然后让万豪的Courtyard酒店采用,我查了一下,万豪旗下有超过2000家Courtyard酒店。

First one is the hotels, we thought we would sell a lot of robots because we were going to go to, say, Marriott, and we'd get Courtyard to do it, I looked it up, and there's over 2,000 Courtyards by Marriott.

Speaker 1

所以我们会每家万豪酒店都部署一台机器人,也许三台,我不确定他们到底需要多少台。

So we'll put a robot in every Marriott, maybe three, I don't know how many they're going to need.

Speaker 1

这就是我向风险投资人们讲述的故事。

This is the story I was telling to the VCs.

Speaker 1

我们遇到了一位副总裁,他说这令人兴奋。

We got this VP who's saying that this is exciting.

Speaker 1

我们面对的是一个庞大的市场,你看看有多少家酒店。

We've got this market with, you know, look at how many hotels there are.

Speaker 1

美国有五万家酒店。

There's 50,000 hotels in The US.

Speaker 1

你可以自己算一算。

You can do the math.

Speaker 1

如果你假设每家酒店用三台机器人,那就是一个十亿美元的市场,这就是答案。

If you assume that you're gonna get three per hotel, it's billion dollar market, that was the answer.

Speaker 1

你总是得从一个十亿美元的市场开始。

You always have to start with a billion dollar market.

Speaker 1

但结果发现,实际情况并不是这样。

But it turned out that it didn't quite work that way.

Speaker 1

万豪、希尔顿和喜达屋并不会为他们的酒店采购机器人。

Marriott and Hilton and Hyatt aren't buy robots for their hotels.

Speaker 1

实际上,他们拥有的酒店数量并没有那么多。

They actually don't have that many hotels.

Speaker 1

他们看起来拥有大量酒店。

They seem to have a lot of hotels.

Speaker 1

他们的名字出现在很多酒店上。

They're named on a lot of hotels.

Speaker 1

他们是特许经营商。

They're franchisors.

Speaker 1

对吧?

Right?

Speaker 1

酒店的实际所有者是各种各样的房地产团体,遍布各地的建筑业主。

The people who own the hotels are building owners, all different little real estate groups all over the place.

Speaker 1

建筑业主会去万豪总部,上四楼,查看万豪旗下所有的品牌,然后决定:我们要做庭院酒店,我们要做某某品牌。

Building owners go to the headquarters of Marriott and they go to the 4th Floor and they look at all the different brands that Marriott has and they pick, well, we're gonna be a courtyard, we're gonna be a whatever.

Speaker 1

万豪的人不会买你的机器人,购买方会是这些业主。

And the Marriott guys are not gonna buy your robots, it's gonna be these owners.

Speaker 1

这意味着对我们来说,不能向万豪销售,必须直接向建筑业主销售。

Well, this means for us, we can't sell to Marriott, we have to sell to building owners.

Speaker 1

所以我们不得不挨家挨户去酒店推销,最终每次只能卖出一台机器人。

So we have to go door to door to hotels and sell one it ended up being one robot at a time.

Speaker 1

我们实际部署后发现,真正需要配送服务的人比我们预期的少得多。

The other thing we found out once we actually deployed them was how many people were actually going to want deliveries, and it wasn't as many as we had hoped.

Speaker 1

在美国,使用率最终只有大约15%。

It ended up being like 15% in The US.

Speaker 1

但在中国,情况却完全不一样。

Now somehow, this is completely different in China.

Speaker 1

如果你去中国,每家酒店都有机器人,就跟我们的一样。

If you go to China, every hotel has robots, just like ours.

Speaker 1

我们在早期并没有和他们达成任何合作。

And we didn't do a deal in the early days with them.

Speaker 1

我们本应该这么做的。

We probably should have.

Speaker 1

好吧,关于这一点我就不多说了,但电梯的市场结构也很重要,你必须和它们打交道。

Okay, so I won't say much more about this except that there's also a market structure of elevators and you have to work with them.

Speaker 1

电梯的一个挑战是,电梯实际上是建筑物中移动机器人的天然敌人,对吧?

One challenge with elevators, elevators And is the natural enemy of mobile robots in buildings, right?

Speaker 1

如果你想要从A点到B点,很可能你需要换楼层。

If you want to go from point A to point B, chances are you have to go to another floor.

Speaker 1

你得乘坐电梯。

You have to ride the elevator.

Speaker 1

当机器人进入电梯,电梯门关闭后,你正在与云端服务通信,而云端服务也在与电梯系统通信,这时你有时会失去与电梯或机器人的连接。

And when the robot gets inside the elevator and the doors close and you're talking to your cloud service that's also talking to the elevator system, you've just lost connection to the elevator sometimes, to the robot.

Speaker 1

机器人现在与外界断开了连接。

The robot's now disconnected from the world.

Speaker 1

那么,机器人怎么知道什么时候该下电梯呢?

So how's the robot gonna know when to get off the elevator?

Speaker 1

系统会向它发送一条消息,比如:你现在在三楼。

Well, the system's gonna tell it a message, like you're on the 3rd Floor now.

Speaker 1

但由于你身处封闭的电梯内,这条消息无法传达到。

And the message doesn't get through because you're in this closed elevator.

Speaker 1

然后门打开了,网络重新连接,门开始关闭。

Then the door opens, and the network reconnects and the door starts to close.

Speaker 1

机器人收到了消息:‘你现在在三楼’,而门已经关上了。

And the robot gets the message, You're on the 3rd Floor now, and the doors are closed.

Speaker 1

机器人看了看,说:‘门还没开呢。’

And the robot looks and goes, Well, door's not open yet.

Speaker 1

我会等。

I'll wait.

Speaker 1

然后电梯到达五楼,门打开,机器人出去后前往305室,而305恰好和505是同一个位置,因为所有楼层的布局都一样。

And then the car gets to the 5th Floor and it opens up, robot goes out and goes to 305, which happens to be the same as 505 because all the floor plans are the same.

Speaker 1

它不会感到困惑。

It doesn't get confused.

Speaker 1

但它现在却在打电话给305,站在505的门前,这就出问题了。

But now it's calling 305 on the phones and standing in front of the door at 505, and you have a problem.

Speaker 1

所以我们不得不解决许多类似这样的技术问题。

So we end up having to solve a lot of technical problems like that.

Speaker 1

在这种情况下,我们最终通过气压计解决了这个问题,机器人通过感知空气压力来判断高度。

In this case, we end up solving it with a barometer, the robot actually sensing the air pressure to know what height.

Speaker 1

你会问,这管用吗?

And you say, does that work?

Speaker 1

实际上,是的,出人意料地有效。

Actually, yeah, surprisingly.

Speaker 1

最终,我们对处理电梯的问题感到厌倦,于是造出了一台能自己按电梯按钮的机器人。

Eventually, we got so fed up with trying to deal with the elevators that we built a robot that could push the elevator buttons itself.

Speaker 1

这是一个聪明的点子,但同时也是一个问题。

And this is a clever idea, and at the same time, it's a problem.

Speaker 1

之所以是个问题,是因为当你乘坐电梯时,其实存在一种社交情境,对吧?

And the reason it's a problem is if you think about going in an elevator, there's a social situation going on, right?

Speaker 1

因此,我们设计了第一个版本:机器人进入电梯后,站立着面向门,安静地等待,到该下车时就准确地离开。

And so, we designed the first version where the robot would go inside, stand, face the door, very quietly, wait for its time to get off, and then it would get off at the right time.

Speaker 1

就这样。

And that was it.

Speaker 1

这根本不会让你感到困扰。

It didn't bother you at all.

Speaker 1

但这个机器人因为需要按按钮,无法在电梯里静止站立。

But this robot, because we had to push the buttons, couldn't stand still inside the elevator.

Speaker 1

它会在里面来回移动。

It's going to be moving around.

Speaker 1

所以你和机器人一起在电梯里,你试图按按钮,它也在试图按按钮。

So you're in the elevator with the robot, which is also, you're trying to push the button, it's also trying to push the button.

Speaker 1

所以并不是它完全不行。

So it's not that it didn't work.

Speaker 1

我们实际上成功地在12家酒店部署了它。

We actually deployed this in 12 hotels successfully.

Speaker 1

但人们并不太喜欢它,对吧?

But people didn't like it as much, right?

Speaker 1

如果你收到投诉,最终就得把产品撤下来。

And then if you get complaints, you end up taking out your product.

Speaker 1

也许关于这件事,我就说到这儿了。

Maybe that's the last I'll say about that.

Speaker 1

最后,为了节省时间,我再讲一句话,然后回答大家的问题。

The last thing I'll just say in the interest of time and then I'll take any questions.

Speaker 1

所以我回到斯坦福来负责这里的机器人中心,作为创业者,这有点不一样,对吧?

So I came back to Stanford to run the robotics center here, which is, as an entrepreneur, is sort of different, right?

Speaker 1

但这也轻松多了。

But it's also much less stressful.

Speaker 1

而且还能让我重新回到机器人研究的前沿,而不是一直处在销售模式中。

And also lets me be back at the cutting edge of robotics instead of having to be in sales mode all the time.

Speaker 1

对我来说,这是一个不错的选择。

And for me, that's a good choice.

Speaker 1

当然,并不是每个人都会有这种感受,但我很高兴能回来做这件事。

Now, not everybody is going to feel that way, but I'm happy to be back doing this.

Speaker 1

机器人中心的目标是汇聚来自各个学科,尤其是工程学院的教授们,共同创造技术,帮助地球上的人们。

The Robotics Center is about bringing together faculty from all over different disciplines, especially in the engineering school, to create stuff, technology, to help people in the planet.

Speaker 1

但我只想再留下最后一句话。

But I'll just leave you with this last thing.

Speaker 1

我们有一个想法,认为这些是我们专注的几个重点领域,我觉得这很可笑,因为你不可能真正同时专注五件事。

We have this idea of these are these flagship areas that we're focused on, which I think is hilarious because you can't really focus on five things.

Speaker 1

你只能专注一件事。

You focus on one thing.

Speaker 1

但我们有很多教授,他们各自专注于不同的领域。

But we have a lot of faculty, they're all focused on different ones.

Speaker 1

机器人学范围很广,但所有领域都共享一个共同的核心。

Robotics is broad and there is this common core that supports them all.

Speaker 1

因此,我很高兴能回来这里从事机器人相关的工作。

So I'm happy to be back here working on robotics.

Speaker 1

我最近经常与试图创办机器人公司的创业者交谈,我总会问第一个问题:你们的市场是什么?

I end up talking to a lot of entrepreneurs who are trying to do robotics companies these days, and I always ask the first question, which is, what's your market?

Speaker 1

他们会告诉我:这是我的三个市场,而我就会问:那到底是哪一个?

And they'll tell me, these are my three markets, and I'll say, which one?

Speaker 1

先弄清楚这一点。

Figure that out first.

Speaker 1

我们真的努力成为连接斯坦福这里各种精彩机器人研究与现实世界应用的桥梁。

And we really are trying to be a bridge between all kinds of cool robotic stuff that's happening here at Stanford and getting things out in the real world.

Speaker 1

因此,我就说到这里,现在开放提问,好,接下来请大家提问。

So with that, let me stop and take questions and, yeah, go from there.

Speaker 0

这太棒了。

That's awesome.

Speaker 0

谢谢你,史蒂夫。

Thank you, Steve.

Speaker 0

我们现在开放提问。

We're gonna we're gonna open up for questions.

Speaker 0

我们大概有十个问题。

We've got about 10.

Speaker 0

我想接着史蒂夫刚才的评论说一下,现在每个人都在创办物理AI公司,这确实是一个备受关注的热门领域。

I wanna piggyback off of that last comment, Steve, about how everybody's now starting a physical AI is a really hot space that people are putting a lot of attention into.

Speaker 0

班级里一个典型的问题是:如果你克隆了自己,并让克隆体去创办一家物理AI公司,你会让它专注于哪个领域?

One of the signature questions from the class is that if you cloned yourself and you can direct your clone to go start a physical AI startup, what area would you have it focusing on?

Speaker 1

我对老年护理机器人充满热情。

I'm totally passionate about senior care robotics.

Speaker 1

我觉得从人口结构来看,未来二十年这显然是一个增长领域。

I feel like the demographics make this an obvious growth area for the next two decades.

Speaker 1

而且我们缺乏足够的人手来照顾所有老年人,所以如果我现在克隆自己,也许我会决定再创办一家公司,但如果真的这么做,我会选择

And we don't have enough people to care for all the older adults, so if I clone myself right now, maybe I'll decide to do another startup, but if I do, it'll be

Speaker 0

这个领域。

in that space.

Speaker 0

好的,太棒了。

Okay, awesome.

Speaker 0

谢谢你,Steve。

Thank you, Steve.

Speaker 0

Tiger。

Tiger.

Speaker 2

好的。

Okay.

Speaker 2

是的,非常感谢你的演讲。

Yeah, thank you so much for your talk.

Speaker 2

我觉得这非常有趣。

I think it's really interesting.

Speaker 2

我的问题是,ROS统一机器人软件,但如今机器人中的AI和自主性发展迅速。

My question is, Ross Unified Robotics Software, but AI and autonomy in robots is moving fast nowadays.

Speaker 2

如果你现在要为未来二十年的机器人重新设计ROS,你会做出哪些改变或不同的选择?

If you had to design Ross all over again right now for the next twenty years of robotics, what would you change or do differently?

Speaker 1

实际上,ROS正在开展一个项目,研究如何集成AI技术。

So actually, ROS is working on there's a project within ROS looking at how to integrate AI technologies.

Speaker 1

技术。

Technologies.

Speaker 1

我认为这取决于你对AI的信仰程度。

I think it depends on how you or say what kind of a believer you are in AI.

Speaker 1

有一类坚定的信仰者会说,把传感器数据输入进去,大量训练,然后输出动作,其他都不用管。

There's the true believers who would say, sensor data in, train a lot, and actions out, and don't worry about anything else.

Speaker 1

而我认为,机器人专家会说,实际上这里存在一个层次结构。

And then I think there's the roboticists who would say, actually, there's a hierarchy there.

Speaker 1

你希望在合适的地方使用机器学习,但并不一定非要以相同的频率处理所有事情。

You want to use machine learning where it's appropriate, but you don't want to necessarily do everything at the same let's say at the same frequency.

Speaker 1

对吧?

Right?

Speaker 1

想想看,摄像头数据是30赫兹的。

Because think about you get camera data at 30 hertz.

Speaker 1

对吧?

Right?

Speaker 1

但当你驱动电机时,你希望达到1000赫兹的频率。

But you want a thousand hertz when you're actually driving motors.

Speaker 1

那么,你真的想让所有东西都以1000赫兹运行吗?这真的能实现吗?

And so do you wanna drive everything at a thousand hertz, and can you actually make that work?

Speaker 1

所以我认为,分层的方法是合适的。

So a hierarchical approach, I think, is appropriate.

Speaker 1

而ROS实际上正是基于这种结构的。

And ROS actually works with that structure.

Speaker 1

因此,你会看到各种ROS包,比如OpenCV之类的,会被训练好的感知系统所取代,或者说被训练好的策略所取代,甚至可能被更高层的规划器取代。

And so I think what you'll see is that various ROS packages will be replaced, let's say OpenCV kind of things, will be replaced by perception systems that are trained policies, if you will, and maybe even higher level planners.

Speaker 1

但它们会以模块的形式存在,而不是整个系统都这样。

But they'll be in points as opposed to the whole thing.

Speaker 2

很好。

Great.

Speaker 2

谢谢。

Thank you.

Speaker 2

谢谢。

Thank you.

Speaker 0

如果其他人有任何问题,可以在这里排队提问。

And if anybody else has any questions, you can queue up here.

Speaker 3

你好。

Hi.

Speaker 3

我想问问你刚才提到的研究与实际应用之间的关系。

I wanted to ask about your last point about research and real world application.

Speaker 3

所以我觉得在斯坦福,我们经常听到关于研究如何能应用于现实世界、惠及更广泛群体的讨论。

So I feel like at Stanford, we hear a lot about how research can become applicable in the real world for the wider community.

Speaker 3

但你能谈谈从学术界转向你刚才提到的实际应用时,会遇到哪些挑战或困难吗?

But can you talk a little bit about the challenges or struggles that comes with trying to incorporate from the academic world to actual real world applications like you were talking about?

Speaker 1

是的,这是个很好的问题。

Yeah, that's a great question.

Speaker 1

当你在做学术研究时,无论是博士生、博士后还是教授,你的目标都是发表论文。

When you're doing academic research, if you're a PhD student, if you're a postdoc, a professor, your goal is to publish.

Speaker 1

有时候我们会不自觉地追求论文数量,或者不断构思下一个重大想法。

Sometimes we find ourselves counting papers or coming up with the next big idea.

Speaker 1

因此,目标有时就是尝试下一个新点子,一旦尝试过,就必须继续前进。

And so the goal sometimes is let me try out the next thing and then having tried it out, need to move on.

Speaker 1

如果你是一家公司,想开发一个产品,就可以反复去做。

And if you're a company and you're trying to do a product, you can do something over again.

Speaker 1

以机器人的导航为例。

So take navigation of robots.

Speaker 1

当我们启动车库轮项目时,导航已经是一个被解决的问题。

When we started Wheel of the Garage, navigation was a solved problem.

Speaker 1

但我们又重新解决了一遍,并把它应用到了Ross One上。

And then we solved it again and we put it into Ross one.

Speaker 1

当时效果还行。

And it was okay.

Speaker 1

当我们启动Savvy Oak时,我们意识到,仅仅把导航技术用在酒店里是不够的。

When we started Savvy Oak, we said, oh, it's not good enough to actually put it out in hotels.

Speaker 1

我们需要再次解决这个问题。

We need to solve it again.

Speaker 1

所以你要不断改进、持续优化。

And so you keep on improving and improving.

Speaker 1

从技术上讲,机器人可以从A点移动到B点,但当你真正将其投入产品时,一些新问题就变得重要了,比如运行的平稳性以及识别反光地板等——这些在原始论文中并未提及,因为当时并未遇到。

Technically, can get from point A to point B, but new things matter when you actually go out into a product like smoothness and recognizing, recognizing, let's let's say, shiny floors, which wasn't in the original paper because it didn't come up.

Speaker 1

但当你真正进入现实世界时,我们花了很长时间解决这个工程问题,因为我们最初部署的一家医院,每天下午两点阳光会照进房间,正好穿过机器人必须经过的区域,导致机器人完全无法识别环境。

But when you actually go out into real world, we spent a long time as an engineering problem because one of the hospitals that we first deployed in, the sunshine came in at 02:00 in the afternoon and basically ran across this area where the robot had to go, and the robot just couldn't see.

Speaker 1

因此机器人就无法前进,因为它不会前往它看不到的地方。

And so it wouldn't go because it doesn't go where it can't see.

Speaker 1

这并不安全。

That's not safe.

Speaker 1

于是机器人就会卡在那里动弹不得。

And so it would just be stuck.

Speaker 1

所以我们最终不得不为机器人发明了虚拟太阳镜,让它能够完成任务。

And we end up having to invent virtual sunglasses for the robot so that it could do its task.

Speaker 1

否则,我们就得告诉客户:下午两点到四点之间别使用它。

Otherwise, have to tell the customer, just don't use it between two and four.

Speaker 1

其他时间它就没问题。

It'll be fine.

Speaker 1

但这真的不可接受。

But that's not really acceptable.

Speaker 4

谢谢。

Thank you.

Speaker 5

你好。

Hello.

Speaker 5

感谢您的演讲。

Thank you for the presentation.

Speaker 5

我是迪安娜,一名计算机科学专业的硕士生。

I'm Deanna, a master's student studying computer science.

Speaker 5

我的问题是,我觉得您在谈到老年护理机器人时已经部分回答了,但我很好奇,您认为机器人技术将在哪个行业产生最深远的变革?

My question I think you sort of answered it with the senior citizen care robots, but I wonder what industry do you see robotics being the most transformative for?

Speaker 5

比如教育领域,我知道家用机器人更受欢迎,但您认为哪个行业会彻底被机器人技术改变?

So for instance, like education, I know domestic robots are a lot more popular, but what industry do you see just completely transforming to robotics?

Speaker 1

是的,我的意思是,我们还没有完成制造业的变革,但这一过程已经持续了六十年,我认为还会继续下去。

Yeah, so I mean, I think we haven't finished transforming manufacturing, but that's been going on for sixty years and I think will continue.

Speaker 1

过去二十年,我们见证了亚马逊的崛起和物流行业的爆发,实体商店基本上被你可以通过手机下单的仓库所取代。

The last twenty years, we had Amazon and we had logistics taking off, basically brick and mortar stores being replaced by warehouses that you order from your your phone.

Speaker 1

因此,机器人技术在这一领域有着巨大的增长空间。

And so there's that's a huge growth area for robotics.

Speaker 1

我认为下一个领域是广泛的服务业。

I believe the next area is service industry broadly.

Speaker 1

我对家用机器人则持一些怀疑态度。

I'm a little bit more skeptical about home robots.

Speaker 1

如果你认真想想,比如你家里有两个人,或者两个人加一两个孩子,其实家务活并没有多到值得花十万美元买一个类人机器人来完成。

I think if you really think about it and you're, say, two people or two people and one or two kids, there sometimes isn't that much housework to do that you would want to spend $100,000 on a humanoid robot to do it.

Speaker 1

但一旦你把护理需求加入进来,比如家里有患阿尔茨海默病的亲人需要照顾,或者有其他残疾需要帮助,一旦涉及护理,经济逻辑就完全改变了。

But as soon as you put caregiving in the mix, as soon as you say, Ah, I've got somebody who's got Alzheimer's in my family that I need to take care of, or whatever disability they may have that they need help, as soon as you have caregiving, the economics change completely.

Speaker 1

因此,我个人正在关注护理领域的机遇,这些机遇会出现在医院、养老机构,也会出现在家庭中。

So I'm personally looking for caregiving opportunities, and those will happen in hospitals, those will happen in senior care facilities, but they also happen in the home.

Speaker 1

我认为当前的挑战是弄清楚人们究竟需要什么,机器人需要完成哪些具体任务,同时也要了解技术的可行性,并将这些要素结合起来。

The challenge, I think, right now is to figure out what's exactly needed, what do the robots need to do for people, and also what's possible and get those things together.

Speaker 1

而可能性正在迅速变化,你可能不会完全取代人类,但可以成为护理人员的助手,这能大大减轻负担,价值非凡。

And what's possible is changing really fast, but you won't necessarily replace a human, but you might become a caregiver's aid, and that might take a huge load off and be worth a lot of money.

Speaker 5

谢谢。

Thank you.

Speaker 5

谢谢。

Thank you.

Speaker 4

你好。

Hi.

Speaker 4

感谢你的演讲。

Thanks for the talk.

Speaker 4

我叫古斯,曾参与过SpaceROS以及阿耳忒弥斯计划中的其他一些项目。

My name is Gus I've worked with, SpaceROS and a few other things in Artemis.

Speaker 4

我们所从事的工作看起来跟你刚才展示的完全不一样。

And the stuff that we work on kind of looks nothing like what you just showed us.

Speaker 4

嗯。

Uh-huh.

Speaker 4

那么,你们在推向市场或向潜在客户或投资者推介时,需要达到什么样的展示水平呢?

So what's the level of presentability that you kind of need to go to market or try to pitch to potential customers or investors?

Speaker 4

因为我们在机器人领域做的某些工作,说实话,根本没什么可展示的。

Because some of the work that we work on in robotics is really, you know, not very presentable at all.

Speaker 4

你觉得在两者之间,一个合适的中间点在哪里?

Where do you think is, like, a good, in between point?

Speaker 1

你所说的‘可展示’是指什么?

When you say presentable

Speaker 4

意思是它看起来像一个成品,能在酒店里正常运作,而不是仅仅展示底层硬件?

In the sense that it looks like a finished product, it looks reasonably workable in a hotel versus maybe just the hardware itself that's underpinning this?

Speaker 1

是的,这很有趣,要让一个产品面世,究竟需要多少打磨。

Yeah, so it's very interesting how much polish you actually need to get something out.

Speaker 1

我们刚开始的时候,我曾挑战过团队,因为之前我们曾在Willow Garage开过一次会,当时高层领导围坐在一张大桌子旁,我终于明白了他们之间一直在争什么。

We started and I kind of challenged the team because there's another story that we had a meeting at Willow Garage at some point where the senior leadership team was around a big table, and I realized there's this fight going on back and forth, and I finally finally understood what it was.

Speaker 1

一方面是软件团队,他们说我们要快速推进、不断迭代;另一方面是硬件团队,他们坚持要用瀑布式开发模型。

And it was the software people on one side who were saying, we need to move fast, we need to iterate, and there were the hardware people on the other side saying, we need like a waterfall model.

Speaker 1

我意识到,当我刚开始学软件工程时,第一堂课上他们给我们展示了瀑布模型,我当时完全不知道那是什么,也不明白为什么需要它,但他们说,这是错的。

And I realized, like when I started in software engineering, the first thing that first software engineering class, they showed us this waterfall model, which I had no idea what it was or why you would want it, and they said, that's wrong.

Speaker 1

你需要循环。

You need loops.

Speaker 1

对吧?

Right?

Speaker 1

后来,这被称为敏捷开发。

Eventually, called that agile.

Speaker 1

好吧。

Okay.

Speaker 1

但这些硬件工程师仍然相信瀑布模型,对吧?

But this hardware people still believe in the waterfall, right?

Speaker 1

而且他们这么想是有道理的,因为他们不得不如此。

And for good reasons, because they had to.

Speaker 1

所以我们实际上进行了讨论并达成一致:我们每个月尝试做一个新的硬件原型,这是一个合理的路径,只要我不对硬件原型抱有任何期望。

So we actually had this discussion and an agreement that we would try to do a new hardware prototype once a month, that that was a reasonable path, as long as I didn't expect anything out of the hardware prototype.

Speaker 1

到了第五个月,我们做出一个看起来像Relay的原型。

And by month five, we had one that actually looked like Relay.

Speaker 1

我们解决了如何为它制作外壳、如何让它足够美观,以至于我们把这个五个月大的原型带到了四季酒店,它已经足够好了。

We had solved how do you make skins for it, how do you make it presentable enough that we actually put that robot, this five month old prototype, was the one that we took to the 4 Seasons, and it was good enough.

Speaker 1

但它还不足以真正运行。

It wasn't good enough to actually work.

Speaker 1

对吧?

Right?

Speaker 1

它沿着走廊走了一段。

It went down the hall.

Speaker 1

它摇摇晃晃,让人感觉不太舒服,但已经足够用来展示,可以展示它的功能和外形。

It was wobbling and didn't make people feel very good, but it was good enough to show off and you could show the functionality and the form.

Speaker 1

作为原型,这样就可以了。

And as a prototype, that was okay.

Speaker 4

明白了。

Got it.

Speaker 4

谢谢。

Appreciate it.

Speaker 4

谢谢。

Thank you.

Speaker 0

谢谢你的提问。

Thank you for the question.

Speaker 0

很抱歉,Gang,我得打断一下,但你们课后也可以单独向Steve提问。

I have to cut it, unfortunately, Gang, but you guys can ask questions also separately to Steve after class.

Speaker 0

时间快到了,请大家和我一起感谢Steve带来的精彩EPR。

But we're coming up on time, so please join me in thanking Steve for that awesome EPR.

关于 Bayt 播客

Bayt 提供中文+原文双语音频和字幕,帮助你打破语言障碍,轻松听懂全球优质播客。

继续浏览更多播客