Tip:
Highlight text to annotate it
X
[MUSIC PLAYING]
RODNEY BROOKS: I'm going to talk about American
manufacturing--
how we solve for American manufacturing.
American manufacturing is not dead despite what many people
thinking about it.
It's a $2 trillion dollar enterprise.
Chinese manufacturing is $2 trillion.
European manufacturing is $2 trillion.
Japanese manufacturing is $1 trillion.
So that just gives you a size of the scale of manufacturing.
But American manufacturing has, I think, lost its
ways in many ways.
It hasn't had the information technology revolution that
everything else in our lives has had.
We've made it less resilient.
After Fukushima in 2011, 10 days later American
manufacturers started shutting down because the supply chain
went via Japan.
And things were not resilient.
So how do we make it resilient?
We had made American manufacturing efficient by
outsourcing the low end.
We started after the Second World War by going to Japan,
where it was a technological society, economy was in ruins,
low cost labor.
But as the economy came back up, it was no
longer low cost labor.
So we moved.
We [? headed ?] off to Korea--
low cost labor.
The society grew.
People became more aware.
People moved up the value chain.
And it became too expensive.
We moved to Taiwan.
Same thing happened there.
Manufacturing was cheap there originally, but
it became more expensive.
Then we moved into southern China via Hong Kong.
The same sort of thing happened.
By the early '90s, it was too expensive to do sewing for
toys in China--
late '90s.
And so we start moving down to Vietnam in the late '90s.
And there things go up.
And eventually we ran out of places to go do low cost
manufacturing.
But doing low cost manufacturing in the far
distant places has had a bad impact on our innovation,
because the innovation tends to move to those places.
Innovation doesn't happen back here where the innovators are.
So there's some movement happening out there
in the world now.
There's a bunch of people doing different things, coming
together in small groups.
And they're spread out across the country, not just in
Silicon Valley, not just in Boston.
There's stuff happening where there's makers, there's tech
shop, there's all this bottom up stuff.
How do we build stuff ourselves?
How do we make that happen?
And I think that's the core of a great idea.
So there are people out there making stuff.
Some of you might know Chris Anderson up in
the right hand corner.
He just left "Wired" to go to his maker space where he's
building stuff for do-it-yourself drones.
But there's lots of people--
bottom right there is a maker bot, which is a 3D printer.
You see it's made of wood, out of New York City.
There's a lot of people doing this stuff.
They're just bubbling away at stuff.
And it reminds me a lot of something that most of you are
too young to remember.
But I remember Silicon Valley, and around the late '70s the
Homebrew Computer Club led to us here.
That led to what we all have.
It led to the people trying out ideas then venture
capitalists coming in, the whole infrastructure coming in
on top of that.
And we got to where we are today with information
technology.
So that's starting to happen with manufacturing and doing
it yourself, building stuff yourself.
But it needs a bunch of things.
The computer revolution, the IT revolution, didn't just
happen by itself.
There were a bunch of other things that were needed.
There were inventions.
Certainly, we need a lot of inventions.
I'm going to show you some inventions.
But we also need infrastructure and new
business models.
And new business models are an opportunity.
There's lots and lots of opportunity in how we change
manufacturing.
I'm going to talk about inventions first.
Inventions are incredibly important to information
technology--
invention after invention.
There's lots of inventions in the manufacturing space.
The one I've been working on for the last four years is
about low cost labor and how we can't have that sort of
Chinese low cost labor in the US.
People don't want those jobs.
So how do we get around that and bring
manufacturing more locally?
So I've been working on a robot called Baxter.
Baxter is a humanoid robot.
It does look like science fiction robots it
turns out in this case.
And I'm going to show you some video of how Baxter operates.
Here's someone programming it.
And she doesn't know about quaternions it turns out.
Whereas current industrial robots, you have to know about
quaternions to program it to do anything.
She just picks it up.
She moves the hand.
She presses some buttons.
The robot figures out what she means.
Now it's going to do the task.
Oh, and it says, I don't like that kinematics
that you gave me.
So it moves the other arm out of the way.
It optimizes the pathway.
It figures it out.
It's looking a little puzzled, so you get some feedback as
this happens.
Ah, OK.
Now I know what to do.
So the idea--
and here, it's a bit washed out, but this is a graphical
user interface.
It's easy to use as a smartphone.
Here's the robot, doesn't know what to do.
She goes in.
She shows it the objects it's supposed to see.
It learns the objects.
She shows the left arm and the right arm-- different objects,
tells them what to do with it.
And in a matter of minutes, it's now able
to sort these objects.
But it's got inbuilt intelligence.
So it missed.
Damn.
But it's OK.
If figures it out.
It goes and tries again.
It's got error recovery built in and goes off and picks up
the object with its left hand that it's supposed to pick up
and goes and puts it where it's supposed to be.
And everything is force based.
So it feels a force.
It lets go.
And it's safe to be around.
Oh, dear.
If that was a current industrial robot,
she would be dead.
But she's not.
But you program this robot by grabbing it, moving it,
showing it.
You don't need any external language.
You don't need to an external computer.
You just show it what you want to do, interact with the
graphical user interface.
You plug it into other machines.
It interacts with the other machines.
And it does real tasks in real factories.
By the way, it's built in the United States.
It's $22,000.
It's a very low cost robot built here.
And the trick we play is we make it so that the end user
doesn't have to know deep stuff.
So the end user, sort of like in an Ikea catalog, gets to
build a hand for the particular task they want to
do with a kit of parts.
They put the pieces together.
They don't need training on how to build hands.
It's all very graphical.
They build a hand, but now the robot can see the hand itself.
So it's got to know what its own hand looks like.
So we let the robot put it up against the red screen, see
through its camera what the hand looks like, and figure
this stuff out.
We don't ask people to do complicated things.
We let the robot do the complicated stuff.
So this robot's now starting shipping.
They're starting to be in factories.
This is Mildred.
Mildred has worked for 25 years in a plastics factory in
Connecticut.
This is one hour after she first saw
an industrial robot--
first time in her life.
In that one hour, she had learnt to program it.
She had made it do tasks in the factory.
She's getting older as are all our factory workers are
getting older.
But by giving them a tool that they can interact with and
they can program, it's not like automation is
coming from on high.
They become the robot supervisors.
They get control of what they're doing.
And I think that's very important.
So that's one sort of invention we need.
There's lots of other inventions.
There's a Form Labs 3D printer.
3D printers are great, but we need 3D
printers to be much better.
They have to be able to deal with plastics and
metal at the same time.
They have to be high speed.
3D printers are too slow for bulk manufacturing.
They need to be able to put electronics in
as they print things.
And the killer app, I think, for 3D printers is if they
could build tools, if they could build tooling so you can
have molds for plastics and metal.
On top of 3D printers, you need super CAD, I call it.
CAD that's parametric.
CAD that involves the manufacturing knowledge and
information in the CAD.
It's not just a WYSIWYG, this is what the part looks like.
It's how the part is built, because once you have that,
now designs become much more portable
than they were before.
So lots of invention.
But we also need infrastructure.
In information technology, we've had this infrastructure
that's grown over the last 30 years, which makes it easy to
do a startup, because you've got the network.
You can just plug into the network.
You've got the infrastructure.
But we need different infrastructure for
manufacturing.
And there's all sorts of infrastructure that we have
lost in the US.
This is a little company in Silicon Valley.
I happen to know it, because the upper person there is my
daughter, Alice.
She started a company to build something.
She went on to Kickstarter, as many people do, raised much
more money than she expected, promised she would manufacture
her product in the US.
Now she had thousands of customers, had to manufacture,
tried to manufacture in the US, couldn't do it.
So she and Bettina headed off to China.
This is all in the space of three or four months.
That's where she had to go in order to scale-up
manufacturing.
And one of the critical things was, 3D printers
were not fast enough.
She needed tooling, couldn't get tooling in the US, had to
build tools in China.
That's a common theme.
You have to build tools in China, because there just
isn't that capability in the US.
So how do we help all these people with these great ideas?
We need infrastructure.
We need quick scale-up.
The president of Flextronics just a couple weeks ago said
he would love to have more manufacturing in the US, but
in Asia he can go from zero to a fully-functioning factory in
three months.
He can just put it all together.
Can't do that in the US, can't do that quick scale-up.
We need supply chain dynamism.
People say we can't bring iPhone manufacturing back to
the US, because there's not a supply chain that can support
building 50 million iPhones every three months where the
models change and have that dynamism in supply chain.
There's a lot of infrastructure we need to
build and infrastructure companies.
And then tooling--
critical, critical, critical.
That robot I showed you we built--
$22,000, made in the US, has almost 200 tools.
We could not build the tools in the US.
We had to build the tools in China and bring
them back to the US.
There is not that capability.
But there's also business models.
Business models have changed every two years in the
information technology space for 30 years.
The business models that operate today, you couldn't
even conceive of 30 years ago.
Many of them you couldn't conceive of 10 years ago.
Twitter just bought Bluefin Labs.
Who knew that Twitter analytics was a big idea that
there were going to be lots of companies competing around.
So the business models change over time.
And business models in manufacturing have not changed
for the last 50 years.
So there's an incredible opportunity there.
So here's just one example of changing a business model.
Right now, a product company, you design the product.
You go over to China.
You build it in China.
Then you pay high oil prices to bring it to the US.
It goes out into retail through some channels.
And the retailers get it out to people, whether they're
brick and mortar or internet retailers.
That's how things work now.
With enough invention, maybe things will
change how it works.
Maybe the product companies sell their designs in Super
CAD to the retailers.
The retailers get local manufacturers to bid on this
package that they have, which includes all manufacturing
information.
And the manufacturers build the stuff, and it goes out to
people, again maybe not via a brick and mortar store.
That's an example of a different sort of business
model that can come about with the right tools.
But there's lots and lots of business models
that can come about.
Like we see in information technology, there's all sorts
of business models about plug-n-play equipment, how
that comes together very quickly.
There's business models about supply chain optimization.
There will be lots and lots of-- just like there's lots
and lots of ad analytics companies-- lots and lots of
companies about supply chain optimization.
And the very nature of how you buy stuff
in your supply chain--
I liken it to GE used to sell jet engines.
They now sell hours of jet engines operating.
Right now, people buy stuff.
Maybe they start buying flow rather than stuff.
So there's all sorts of things that can
happen in business models.
So we've got people out there doing stuff.
We've got these makers and various people doing things.
But if we come together and think about this together, we
can do lots of things and change the way
manufacturing works.
And so we see people, makers.
And there's one of the makers here in the front row, Steve.
You know that guy.
So Steve in his weekends, he makes stuff.
And he carries that over to his weekday job and invests in
companies that make bigger versions of that stuff.
So these makers are doing stuff.
There's people building 3D printers.
There's people doing things.
As we come together, we can make them much more efficient,
because they build some radical stuff.
But sometimes it's a little too much Steve Wozniak and not
enough Steve Jobs.
And we need to help this movement with a
little more of that.
And so my call to action here is all of us, let's start
building stuff and let's make the stuff.
Let's make it in the US.
We can do it.
And we get a much more resilient manufacturing.
Thank you.
[APPLAUSE]
[MUSIC PLAYING]