A camera that can see around corners | David Lindell

Transcriber: Ivana Korom Reviewer: Krystian Aparta

譯者: Zoe Chang 審譯者: Helen Chang

In the future,

未來的自動駕駛汽車（自駕車）， 將比人類駕駛的更安全、更可靠。

self-driving cars will be safer and more reliable than humans.

But for this to happen,

在那之前，

we need technologies that allow cars to respond

我們需要發展新的科技，

確保車子的反應速度比人類快，

faster than humans,

we need algorithms that can drive better than humans

其演算法的駕駛技術比人類好，

and we need cameras that can see more than humans can see.

還配備著視力 勝於人類肉眼的攝像鏡頭。

For example, imagine a self-driving car is about to make a blind turn,

例如，想像一輛自駕車即將要轉彎，

and there's an oncoming car

視線範圍外的路上 有輛車迎面而來，

or perhaps there's a child about to run into the street.

或是有小孩跑向馬路中間。

Fortunately, our future car will have this superpower,

還好，未來的車將擁有某種超能力──

a camera that can see around corners to detect these potential hazards.

配備著能偵測出視線外 潛在危機的攝像機。

For the past few years as a PhD student

我是博士生，過去這幾年，

in the Stanford Computational Imaging Lab,

我在史丹佛的電腦成像實驗室，

I've been working on a camera that can do just this --

研究這具有特殊能力的攝像機──

a camera that can image objects hidden around corners

能透視彎角的攝像機，

它能顯示視線正前方 被擋住而看不見的物體。

or blocked from direct line of sight.

So let me give you an example of what our camera can see.

讓我舉例說明這個攝像機的功能。

This is an outdoor experiment we conducted

這是一個我們做的戶外實驗，

where our camera system is scanning the side of this building with a laser,

這個攝像系統 用雷射掃描建築物的側面，

and the scene that we want to capture

而我們要捕捉的景象

is hidden around the corner behind this curtain.

則藏在布幕的後方。

So our camera system can't actually see it directly.

所以攝像系統其實無法直接拍到。

And yet, somehow,

可是，

our camera can still capture the 3D geometry of this scene.

它卻能拍出到布簾後方 物件的立體形狀。

So how do we do this?

這是怎麼辦到的？

The magic happens here in this camera system.

魔法就藏在這個攝像系統裡。

You can think of this as a type of high-speed camera.

你可以把它想成一種高速攝像機。

Not one that operates at 1,000 frames per second,

拍攝速度不是每秒一千張，

or even a million frames per second,

也不是每秒一百萬張，

but a trillion frames per second.

而是每秒一兆張之多。

So fast that it can actually capture the movement of light itself.

速度之快可以捕抓到光影。

And to give you an example of just how fast light travels,

讓我舉例說明光線移動的速度，

let's compare it to the speed of a fast-running comic book superhero

以漫畫中移動速度極快的超人來說，

who can move at up to three times the speed of sound.

超人號稱可以音速三倍的速度移動。

It takes a pulse of light about 3.3 billionths of a second,

可是光脈衝只需大約 33 億分之一秒，

or 3.3 nanoseconds,

或者說 3.3 奈秒，

to travel the distance of a meter.

就能行進一公尺的距離。

Well, in that same time,

也就是說，在同樣的時間內，

our superhero has moved less than the width of a human hair.

超人移動的距離還少於 人類一根頭髮的直徑。

That's pretty fast.

那夠快了吧。

But actually, we need to image much faster

可是，想要拍攝光線 移動不到一公分距離的畫面，

if we want to capture light moving at subcentimeter scales.

速度得要更快。

So our camera system can capture photons

這個攝像系統可以捕捉到光子，

at time frames of just 50 trillionths of a second,

因為拍攝速度達到每張 只需 50 兆之一秒，

or 50 picoseconds.

或者說 50 皮秒。

So we take this ultra-high-speed camera

我們將這個超高速攝像機，

and we pair it with a laser that sends out short pulses of light.

與一台能夠發射超短 光脈衝的雷射機搭配。

Each pulse travels to this visible wall

將光脈衝發射到這面牆壁上，

and some light scatters back to our camera,

有一部分的光會反射回攝像機來，

but we also use the wall to scatter light around the corner

牆面也同時將光線散射到彎角後方，

to the hidden object and back.

碰到隱藏物，再反射回來。

We repeat this measurement many times

多次重複操作後

to capture the arrival times of many photons

就能收集牆面不同區域 光子的抵達時間，

from different locations on the wall.

And after we capture these measurements, we can create

我們用收集到的所有數據，

把牆面上的光影拍成 每秒一兆張影像的影片。

a trillion-frame-per-second video of the wall.

While this wall may look ordinary to our own eyes,

以人類肉眼來看，牆就是一面牆，

at a trillion frames per second, we can see something truly incredible.

而每秒播放一兆張的影片 卻能顯示出肉眼看不見的東西。

We can actually see waves of light scattered back from the hidden scene

甚至能看到視線外的物件 散射回來的光波，

and splashing against the wall.

像海浪般潑濺在牆上。

And each of these waves carries information

每個光波都帶有該隱藏物的訊息。

about the hidden object that sent it.

So we can take these measurements

我們收集這些數據資料，

and pass them into a reconstruction algorithm

再用影像重建演算法

to then recover the 3D geometry of this hidden scene.

畫出該隱藏物件的 3D 幾何形狀。

Now I want to show you one more example of an indoor scene that we captured,

接下來是我們做的另一個 室內場景的範例，

this time with a variety of different hidden objects.

這次用到更多不同的物件。

And these objects have different appearances,

各自有不同的外觀，

so they reflect light differently.

反射光的方式也都不同。

For example, this glossy dragon statue reflects light differently

例如，這個有光澤的龍雕像反射的光

than the mirror disco ball

不同於那個迪斯可的鏡球吊燈，

or the white discus thrower statue.

也跟那個白色鐵餅投手雕像不同。

And we can actually see the differences in the reflected light

從它們各自反射的光， 也能看出其中的差異，

by visualizing it as this 3D volume,

只要將剛剛拍攝的 影片窗格堆疊起來，

where we've just taken the video frames and stacked them together.

就能做出這樣一個 視覺化的 3D 影像。

And time here is represented as the depth dimension of this cube.

立方體的深度代表時間軸。

These bright dots that you see are reflections of light

你看到的這些較亮的光點

是舞廳球燈上的小鏡面反射的光

from each of the mirrored facets of the disco ball,

scattering against the wall over time.

隨著時間的推移射到牆面上。

The bright streaks of light that you see arriving soonest in time

那些最明亮的光束， 也是最快反射到牆面的，

are from the glossy dragon statue that's closest to the wall,

則是來自最靠近牆壁的亮面龍雕像，

and the other streaks of light come from reflections of light from the bookcase

還有其他的光線則是從書櫃 及投手雕像反射回來的。

and from the statue.

Now, we can also visualize these measurements frame by frame,

我們也可以把所收集到的數據

用影像的方式一格一格地播放，

as a video,

to directly see the scattered light.

就能看到四散飛濺的光。

And again, here we see, first, reflections of light from the dragon,

最先看到的還是來自龍反射的光，

closest to the wall,

因為它離牆壁最近，

followed by bright dots from the disco ball

接下來就是來自迪斯可球鏡的光點，

and other reflections from the bookcase.

再來是書櫃，

最後看到的是反射自雕像的光波。

And finally, we see the reflected waves of light from the statue.

These waves of light illuminating the wall

這些光波照亮了牆壁，

are like fireworks that last for just trillionths of a second.

像煙火那樣，只持續了幾兆分之一秒。

And even though these objects reflect light differently,

雖然各物體反射光的方式不同，

we can still reconstruct their shapes.

我們仍能重建它們的形狀，

And this is what you can see from around the corner.

看到角落後面是這個樣子。

Now, I want to show you one more example that's slightly different.

讓我再展示另一個稍微不同的例子。

In this video, you see me dressed in this reflective suit

影片中的我身上穿著反光服，

and our camera system is scanning the wall at a rate of four times every second.

攝像系統每秒掃描牆壁四次。

The suit is reflective,

因為反光服的布料特性，

so we can actually capture enough photons

我們得以捕捉到足夠的光子，

即使攝像機並非正對著我， 仍能拍到我的位置及動作。

that we can see where I am and what I'm doing,

without the camera actually directly imaging me.

By capturing photons that scatter from the wall to my tracksuit,

光子從牆壁彈射到反光服上， 再從反光服彈回牆上，

back to the wall and back to the camera,

再從牆壁射到攝像機，

we can capture this indirect video in real time.

這些非直接的影像都是 以實時攝影捕捉的。

And we think that this type of practical non-line-of-sight imaging

我們認為這類實用的 「非視線」成像技術

could be useful for applications including for self-driving cars,

除了用在自駕車上，

but also for biomedical imaging,

還可用於生物醫學成像，

where we need to see into the tiny structures of the body.

拍攝身體中極其微小的結構。

And perhaps we could also put similar camera systems on the robots

甚至能將類似的攝像系統 裝在機器人上，

that we send to explore other planets.

送到外太空去探索其他星球。

Now you may have heard about seeing around corners before,

或許你們之前已經聽過 透視轉角的技術，

but what I showed you today would have been impossible

但我今天示範的 這兩年才被開發出來。

just two years ago.

For example, we can now image large, room-sized hidden scenes outdoors

例如我們現在可以在室外

實時拍攝房間大小的隱藏場景。

and at real-time rates,

and we've made significant advancements towards making this a practical technology

我們取得的長足進步 使其成為實用的技術，

that you could actually see on a car someday.

可被用在汽車上。

But of course, there's still challenges remaining.

當然目前仍有些挑戰。

For example, can we image hidden scenes at long distances

例如一旦距離拉大，

where we're collecting very, very few photons,

目前使用不傷眼的低功率雷射

所收集到的光子量可能不足以成像？

with lasers that are low-power and that are eye-safe.

Or can we create images from photons

那些不僅碰撞轉角一次，

that have scattered around many more times

而是多次反彈的光子，

還能被用來呈像嗎？

than just a single bounce around the corner?

Can we take our prototype system that's, well, currently large and bulky,

還有，目前這套 系統原型龐大又笨重，

and miniaturize it into something that could be useful

要怎樣將之縮小，

for biomedical imaging

用在生物醫學成像 或家庭保全系統呢？

or perhaps a sort of improved home-security system,

or can we take this new imaging modality and use it for other applications?

除了這些，還有其他的應用方式嗎？

I think it's an exciting new technology

我認為這是一項令人興奮的新技術，

and there could be other things that we haven't thought of yet

應該還有其他的應用方式 我們尚未想到。

to use it for.

And so, well, a future with self-driving cars

自駕車普及的未來 或許現今看似遙遠，

may seem distant to us now --

we're already developing the technologies

但我們正開發讓汽車 更安全、更智慧的技術。

that could make cars safer and more intelligent.

And with the rapid pace of scientific discovery and innovation,

科技創新日新月異，

you never know what new and exciting capabilities

令人驚艷的新功能或許指日可待。

could be just around the corner.

(Applause)

（掌聲）