Later that year, he took it out for the first public test drive,

年末，它執行了第一次路駕測試。

and -- true story -- crashed into a wall.

而最後，撞毀了。

For the last 130 years,

在這130年裡，

we've been working around that least reliable part of the car, the driver.

我們在汽車的發展都圍繞於汽車中最不可靠的部分，駕駛。

We've made the car stronger.

我們加強了車架強度。

We've added seat belts, we've added air bags,

加入了安全帶，和安全氣囊。

and in the last decade, we've actually started trying to make the car smarter

而在近十年，我們終於開始想辦法讓車子更聰明，

to fix that bug, the driver.

這樣才能修復真正的問題，駕駛的存在。

Now, today I'm going to talk to you a little bit about the difference

今天我要談談兩種汽車的不同，分別是：

between patching around the problem with driver assistance systems

為駕駛而加上的附加設施所做出來的車，

and actually having fully self-driving cars

和真正的無人駕駛車。

and what they can do for the world.

它們會在世界上產生什麼影響。

I'm also going to talk to you a little bit about our car

我也要談一下我自家的車子，

and allow you to see how it sees the world and how it reacts and what it does,

讓你們看看它是如何認識這個世界，以及對周遭事物做出反應。

but first I'm going to talk a little bit about the problem.

但首先，我想先談談我們所遭遇的問題。

And it's a big problem:

而這是一項重大的問題：

1.2 million people are killed on the world's roads every year.

每年有120萬人死於交通事故。

In America alone, 33,000 people are killed each year.

單在美國，每年就有33000人。

To put that in perspective,

把它跟其他東西比較一下：

that's the same as a 737 falling out of the sky every working day.

這跟每天都掉一架載滿人的波音737一樣多。

It's kind of unbelievable.

還真難以置信。

Cars are sold to us like this,

車子以這種形式賣給我們，

but really, this is what driving's like.

但實際上，真正駕駛在開車時，卻是這副模樣。

Right? It's not sunny, it's rainy,

對吧？不是大晴天，是雨天，

and you want to do anything other than drive.

你想開車之餘還能打發時間。

And the reason why is this:

為何會造成這樣的狀況是因為這個：

Traffic is getting worse.

交通情況更惡化了。

In America, between 1990 and 2010,

在美國，1990年到2010年，

the vehicle miles traveled increased by 38 percent.

交通工具所行使的英哩數成長了38%。

We grew by six percent of roads,

而每年卻只多增加了6%的道路，

so it's not in your brains.

所以這不是你的錯覺。

Traffic really is substantially worse than it was not very long ago.

交通在近幾年來真的惡化了不少，

And all of this has a very human cost.

而這完全是跟人有關。

So if you take the average commute time in America, which is about 50 minutes,

如果你拿普通美國人的通勤時間，大約50分鐘來計算，

you multiply that by the 120 million workers we have,

去乘上全美一億兩千萬的工作人口，

that turns out to be about six billion minutes

那將會得到大約60億分鐘，

wasted in commuting every day.

每天浪費在通勤上的時間。

Now, that's a big number, so let's put it in perspective.

那是個很大的數字，所以我們把它跟其他東西做個比較：

You take that six billion minutes

你拿這60億分鐘

and you divide it by the average life expectancy of a person,

去除上每人的平均壽命，

that turns out to be 162 lifetimes

那大約是162人一生的時間

spent every day, wasted,

全部浪費在通勤上。

just getting from A to B.

就只為了從A點到B點而已。

It's unbelievable.

難以置信。

And then, there are those of us who don't have the privilege of sitting in traffic.

另外，也有不少交通上的弱勢。

So this is Steve.

這是Steve。

He's an incredibly capable guy,

他是個健全的人，

but he just happens to be blind,

除了他是盲人這一點。

and that means instead of a 30-minute drive to work in the morning,

而這代表著他每天早上不是花30分鐘開車去上班，

it's a two-hour ordeal of piecing together bits of public transit

而是要經過2小時，搭乘各種大眾運輸工具，

or asking friends and family for a ride.

或請朋友或家人帶他去上班。

He doesn't have that same freedom that you and I have to get around.

他沒有你我的自由。

We should do something about that.

我們應該要為他做點什麼。

Now, conventional wisdom would say

一般大眾會告訴你，

that we'll just take these driver assistance systems

我們只要拿這套輔助駕駛的系統，

and we'll kind of push them and incrementally improve them,

然後盡力的去增加他的配備就好了。

and over time, they'll turn into self-driving cars.

到最後，這就會變成了無人駕駛車。

Well, I'm here to tell you that's like me saying

這就好像我跟你說：

that if I work really hard at jumping, one day I'll be able to fly.

如果我很努力的練習跳躍，最後我將能飛。

We actually need to do something a little different.

但實際上，我們必須要做些不同的事才能達到那樣的境界。

And so I'm going to talk to you about three different ways

我接下來要告訴你：

that self-driving systems are different than driver assistance systems.

無人駕駛車跟駕駛輔助系統的3大不同點。

And I'm going to start with some of our own experience.

首先來談談我自己的經驗。

So back in 2013,

2013年，

we had the first test of a self-driving car

我們執行了首次無人駕駛車的測試，

where we let regular people use it.

實驗者是一般大眾。

Well, almost regular -- they were 100 Googlers,

嗯，幾乎啦--他們是100位Google的員工，

but they weren't working on the project.

但這100位員工並不是在這項專案下工作。

And we gave them the car and we allowed them to use it in their daily lives.

我們給他們這些無人駕駛車，並請他們用在他們的日常生活裡。

But unlike a real self-driving car, this one had a big asterisk with it:

但跟實際上開無人駕駛車不同，這次有個前提：

They had to pay attention,

他們必須要專注在車子的自動操作上，

because this was an experimental vehicle.

因為這還只是實驗用的車子而已。

We tested it a lot, but it could still fail.

我們已經做過許多測試，但還是可能失靈。

And so we gave them two hours of training,

我們花了2小時教他們如何使用，

we put them in the car, we let them use it,

請他們做到車內，讓他們實際去測試。

and what we heard back was something awesome,

而我們所聽到的迴響都是好的，

as someone trying to bring a product into the world.

好像他們在為這項產品推銷似的。

Every one of them told us they loved it.

每個人都讚譽有嘉。

In fact, we had a Porsche driver who came in and told us on the first day,

事實上，第一天有位開保時捷的車主跟我們說：

"This is completely stupid. What are we thinking?"

「這太扯了，你們到底在想些什麼？」

But at the end of it, he said, "Not only should I have it,

但到最後，他說：「不只我需要一輛，

everyone else should have it, because people are terrible drivers."

每人都應該要有一輛無人駕駛車，因為人們都不是好駕駛。」

So this was music to our ears,

這對我們來講是無比的鼓勵，

but then we started to look at what the people inside the car were doing,

我們開始去注意人們都在車內做什麼，

and this was eye-opening.

而這才令人大開眼界。

Now, my favorite story is this gentleman

我最喜歡的故事是一位男士，

who looks down at his phone and realizes the battery is low,

他低頭看了一下他的手機，發現手機快沒電了。

so he turns around like this in the car and digs around in his backpack,

結果他轉身到後座的背包找東西，

pulls out his laptop,

然後拿出了一台筆電，

puts it on the seat,

他把筆電放到旁邊的座位上，

goes in the back again,

又再轉身到他的背包裡找東西，

digs around, pulls out the charging cable for his phone,

然後拿出了一條手機充電線，

futzes around, puts it into the laptop, puts it on the phone.

他解開充電線，把它連上手機和筆電。

Sure enough, the phone is charging.

當然，他手機這時在充電了。

All the time he's been doing 65 miles per hour down the freeway.

但別忘了，他在做這些事情時，車子正以時速65英里的狀況下在高速公路上狂奔。

Right? Unbelievable.

難以置信。

So we thought about this and we said, it's kind of obvious, right?

我們稍微想了一下，然後得出一個顯而易見的結論：

The better the technology gets,

當科技越發達，

the less reliable the driver is going to get.

駕駛就越不可靠。

So by just making the cars incrementally smarter,

如果我們只是讓車子很聰明，

we're probably not going to see the wins we really need.

那跟我們所期待的結果將不相同。

Let me talk about something a little technical for a moment here.

讓我稍微談談技術層面的部分：

So we're looking at this graph, and along the bottom

這裡有張圖表，在底部

is how often does the car apply the brakes when it shouldn't.

是代表車子多常在不該煞車的時候煞車。

You can ignore most of that axis,

你可以忽略掉X軸後面的部分，

because if you're driving around town, and the car starts stopping randomly,

因為如果你的車會在你開車時，不停的煞車的話，

you're never going to buy that car.

你大概不會想買這種車子。

And the vertical axis is how often the car is going to apply the brakes

而縱軸是代表車子多常在

when it's supposed to to help you avoid an accident.

該煞車的時候煞車。

Now, if we look at the bottom left corner here,

在左下角這一點，

this is your classic car.

這是代表你有的普通的車子。

It doesn't apply the brakes for you, it doesn't do anything goofy,

他不會幫你踩煞車，或做任何花招，

but it also doesn't get you out of an accident.

但同樣的他也不會防止你發生事故。

Now, if we want to bring a driver assistance system into a car,

如果我們想要將駕駛輔助系統帶到車子裡，

say with collision mitigation braking,

像是碰撞緩解制動系統。

we're going to put some package of technology on there, and that's this curve,

我們需要加裝許多科技在上面。他的曲線長這樣。

and it's going to have some operating properties,

他會有一些操作系統，

but it's never going to avoid all of the accidents,

但沒辦法避開所有事故，

because it doesn't have that capability.

因為它沒有那種能力。

But we'll pick some place along the curve here,

但如果我們在這條曲線上找到一個對的點，

and maybe it avoids half of accidents that the human driver misses,

或許就可以減少一半的事故，相較於只有人類駕駛而言。

and that's amazing, right?

這很驚人，對吧？

We just reduced accidents on our roads by a factor of two.

我們僅靠改變一兩個因素就將事故發生率降低一半。

There are now 17,000 less people dying every year in America.

美國現在每年約有17000人因交通事故而身亡‧

But if we want a self-driving car,

但如果我們想打造一輛無人駕駛車，

we need a technology curve that looks like this.

我們就需要將曲線改變成這樣。

We're going to have to put more sensors in the vehicle,

我們要將更多感應器安裝到車上，

and we'll pick some operating point up here

並將性能調整至大約在曲線上的這一點，

where it basically never gets into a crash.

這樣基本上就能夠避免發生事故。

They'll happen, but very low frequency.

事故仍可能發生，但機率非常低。

Now you and I could look at this and we could argue

現在你我可以去爭辯說

about whether it's incremental, and I could say something like "80-20 rule,"

這條曲線是否有加成性，我可以告訴你說這符合「80/20法則」。

and it's really hard to move up to that new curve.

很難去達到最新的那一條曲線。

But let's look at it from a different direction for a moment.

但讓我們從另一個角度來切入：

So let's look at how often the technology has to do the right thing.

我們來看看這項新技術有多常下出正確的決定。

And so this green dot up here is a driver assistance system.

這個綠點是駕駛輔助系統的表現。

It turns out that human drivers

這顯示出美國駕駛人

make mistakes that lead to traffic accidents

常在約每十萬英里時

about once every 100,000 miles in America.

發生足以造成事故的錯誤‧

In contrast, a self-driving system is probably making decisions

相對的，自駕系統大約

about 10 times per second,

每秒會做10次決定。

so order of magnitude,

在加乘的情況下，

that's about 1,000 times per mile.

大約是每英里會做出1000次決定。

So if you compare the distance between these two,

如果你將這兩點的距離去做比較的話，

it's about 10 to the eighth, right?

這大概有10的8次方(一億倍)的差距吧？

Eight orders of magnitude.

8次方的差距，

That's like comparing how fast I run

這好像拿我跑的速度

to the speed of light.

跟光速來比較。

It doesn't matter how hard I train, I'm never actually going to get there.

這跟我做了多少練習無關，而是我根本就達不到那種境界。

So there's a pretty big gap there.

所以那是一段很大的差距。

And then finally, there's how the system can handle uncertainty.

最後，我們來談談這個系統如何應付突發狀況。

So this pedestrian here might be stepping into the road, might not be.

畫面中這個行人可以算是站在路上，也可以算是站在路旁。

I can't tell, nor can any of our algorithms,

我無法判定，當然我們的程式也無法判定。

but in the case of a driver assistance system,

但如果是駕駛輔助系統要做出反應，

that means it can't take action, because again,

它將沒有辦法反應，因為同樣的，

if it presses the brakes unexpectedly, that's completely unacceptable.

如果它在無預警的情況下煞車的話，那是不能被容忍的錯誤。

Whereas a self-driving system can look at that pedestrian and say,

而當自駕系統看到這位行人時，它會說

I don't know what they're about to do,

「我不知道這人接下來想做什麼，

slow down, take a better look, and then react appropriately after that.

那我就先慢下來，看清楚狀況，然後再視情況做出應變。」

So it can be much safer than a driver assistance system can ever be.

這將比駕駛輔助系統還更安全。

So that's enough about the differences between the two.

這大概就是這兩種系統的差別。

Let's spend some time talking about how the car sees the world.

讓我們花點時間來看看自駕車是如何認識這個世界。

So this is our vehicle.

這是我們的自駕車。

It starts by understanding where it is in the world,

它會先確定自己身在何處，

by taking a map and its sensor data and aligning the two,

靠著結合地圖和感應信息，

and then we layer on top of that what it sees in the moment.

接著我們在這層訊息上加上另一種訊息，

So here, all the purple boxes you can see are other vehicles on the road,

在這裡，所有你看到的紫色盒子都代表著路上的其他車輛，

and the red thing on the side over there is a cyclist,

旁邊紅色的輪廓代表腳踏車，

and up in the distance, if you look really closely,

在上方比較遠的部分，如果你仔細看的話，

you can see some cones.

你會看到一些角椎。

Then we know where the car is in the moment,

當我們知道這些車子的位置的同時，

but we have to do better than that: we have to predict what's going to happen.

我們必須更進一步：去預測車子的動向。

So here the pickup truck in top right is about to make a left lane change

這裡，被標示出來的卡車要切到左線

because the road in front of it is closed,

因為前方的道路封閉了，

so it needs to get out of the way.

它必須要閃開。

Knowing that one pickup truck is great,

知道那輛挑選出來的卡車行徑固然很好，

but we really need to know what everybody's thinking,

但實際上，我們需要知道所有人在想什麼，

so it becomes quite a complicated problem.

這變成了一個很複雜的問題。

And then given that, we can figure out how the car should respond in the moment,

在知道這些情況下，我們可以知道車子該如何應對，

so what trajectory it should follow, how quickly it should slow down or speed up.

它應該跟著哪條路線，它應該多快地去減速和加速。

And then that all turns into just following a path:

這些抉擇就形成了這條路線：

turning the steering wheel left or right, pressing the brake or gas.

該左轉或右轉，踩煞車或油門。

It's really just two numbers at the end of the day.

到最後只是在這兩項做決定而以。

So how hard can it really be?

這有多難？

Back when we started in 2009,

這是當初2009年時

this is what our system looked like.

我們系統的長相。

So you can see our car in the middle and the other boxes on the road,

你可以看到中間是我們的車，以及路上周遭的盒子

driving down the highway.

在高速公路上行駛。

The car needs to understand where it is and roughly where the other vehicles are.

車子需要知道自己的位置及其他車大略的位置。

It's really a geometric understanding of the world.

這其實是用幾何學去了解世界。

Once we started driving on neighborhood and city streets,

當我們開始去模擬在社區或市區裡行駛時，

the problem becomes a whole new level of difficulty.

問題又提升了一個層次。

You see pedestrians crossing in front of us, cars crossing in front of us,

你可以看到路人在我們眼前穿越，車子在我們眼前穿越，

going every which way,

他們往各個方向移動。

the traffic lights, crosswalks.

紅綠燈，斑馬線，

It's an incredibly complicated problem by comparison.

相較於前，這個問題複雜許多。

And then once you have that problem solved,

當你解決這個問題時，

the vehicle has to be able to deal with construction.

車子就可以開始建設路段了。

So here are the cones on the left forcing it to drive to the right,

這裡可以看到，前面有一個角椎，迫使車子必須切到右線，

but not just construction in isolation, of course.

但當然，不僅是單單建構路段，

It has to deal with other people moving through that construction zone as well.

它還得應付如果有人剛好走在在建構中的路段的情況。

And of course, if anyone's breaking the rules, the police are there

當然，如果有人違規了，警察就會到場，

and the car has to understand that that flashing light on the top of the car

車子必須了解車頂在閃的燈號

means that it's not just a car, it's actually a police officer.

代表著它不是一輛普通的車，它是一輛警車。

Similarly, the orange box on the side here,

同樣的，旁邊有個橙色的箱子，

it's a school bus,

這是校車。

and we have to treat that differently as well.

我們也必須對它做出不同的回應。

When we're out on the road, other people have expectations:

當我們開車在路上，有些人會有預期：

So, when a cyclist puts up their arm,

當腳踏車騎士伸出手臂時，

it means they're expecting the car to yield to them and make room for them

它會預期車子會禮讓它，留出空間

to make a lane change.

讓它切換線道。

And when a police officer stood in the road,

當警車在路中間指揮交通時，

our vehicle should understand that this means stop,

我們的車子會了解到，這代表要停車。

and when they signal to go, we should continue.

當看到前進的信號時，我們可以繼續前行。