A headset that reads your brainwaves | Tan Le

譯者: Jeannie Cheng 審譯者: Sunshine Wang

Up until now, our communication with machines

直到現在，我們與機器的溝通

has always been limited

仍局限於

to conscious and direct forms.

有意識和直接的模式

Whether it's something simple

不論是一些簡單的事情

like turning on the lights with a switch,

如用開關開燈

or even as complex as programming robotics,

或一些複雜的程式來控制機械人

we have always had to give a command to a machine,

我們都要給機器輸入一個

or even a series of commands,

甚至一系列的指令

in order for it to do something for us.

才能命令它執行一些動作

Communication between people, on the other hand,

相反的，人與人的溝通

is far more complex and a lot more interesting

就更加複雜和有趣得多

because we take into account

因為我們會考慮到

so much more than what is explicitly expressed.

言語未表達的言外之意

We observe facial expressions, body language,

我們會觀察表情、肢體語言

and we can intuit feelings and emotions

在對話中我們會用直覺來

from our dialogue with one another.

感受對方的感覺和情緒

This actually forms a large part

這些都是做決定時

of our decision-making process.

一些重要的因素

Our vision is to introduce

我們的願景是引進

this whole new realm of human interaction

全新的人與電腦的互動科技

into human-computer interaction

到人類互動的領域

so that computers can understand

這麼一來電腦不只可以

not only what you direct it to do,

明白你指示它所做的事情

but it can also respond

而且也會對面部表情

to your facial expressions

和情緒經歷

and emotional experiences.

作出反應

And what better way to do this

還有什麼比從大腦的

than by interpreting the signals

情感控制中樞直接解譯

naturally produced by our brain,

大腦產生的電波

our center for control and experience.

來得更好呢？

Well, it sounds like a pretty good idea,

這聽起來好像是不錯的主意

but this task, as Bruno mentioned,

但這個任務，正如Bruno所說

isn't an easy one for two main reasons:

並不容易，原因有兩個

First, the detection algorithms.

第一是大腦的偵查演算法

Our brain is made up of

我們的腦是由

billions of active neurons,

數十億個活躍的神經元所組成

around 170,000 km

如果把神經細胞的軸索連在一起

of combined axon length.

大概有十七萬公里

When these neurons interact,

這些神經元互動時

the chemical reaction emits an electrical impulse,

產生的化學作用所發射出的電脈衝

which can be measured.

能夠被測量到

The majority of our functional brain

大部分功能性腦

is distributed over

是分佈在

the outer surface layer of the brain,

大腦的表層

and to increase the area that's available for mental capacity,

心智能力功能也位於此，為了增加表面積

the brain surface is highly folded.

大腦皮質層有非常多的褶皺

Now this cortical folding

大腦皮質褶皺

presents a significant challenge

對分析電脈衝

for interpreting surface electrical impulses.

帶來一個很大的挑戰

Each individual's cortex

每個人大腦皮質層

is folded differently,

的褶皺都不同

very much like a fingerprint.

就像指紋一樣

So even though a signal

因此電脈衝訊息

may come from the same functional part of the brain,

雖然來自功能腦同樣的區域

by the time the structure has been folded,

但大腦皮質褶皺結構早已形成

its physical location

在不同的人的大腦裡

is very different between individuals,

即使是雙胞胎

even identical twins.

訊息發生位置也不同

There is no longer any consistency

大腦皮質層電脈衝訊息

in the surface signals.

沒有一致性

Our breakthrough was to create an algorithm

我們的突破是建立一個演算法

that unfolds the cortex,

攤開大腦皮質層

so that we can map the signals

去勘測這些

closer to its source,

訊息的原點

and therefore making it capable of working across a mass population.

繼而把它運用在大眾身上

The second challenge

第二項挑戰是

is the actual device for observing brainwaves.

觀察腦電波的儀器

EEG measurements typically involve

腦波測量基本上包括

a hairnet with an array of sensors,

一個有許多感應器的髮網

like the one that you can see here in the photo.

就像現在圖中所看到的

A technician will put the electrodes

技術人員會把電極

onto the scalp

用導電的膠或漿糊

using a conductive gel or paste

固定在頭皮上

and usually after a procedure of preparing the scalp

這個準備程序需要在頭皮製造

by light abrasion.

輕微的擦傷

Now this is quite time consuming

這個程序既費時

and isn't the most comfortable process.

又不舒服

And on top of that, these systems

再加上，這些系統

actually cost in the tens of thousands of dollars.

非常昂貴，得花上數萬美金

So with that, I'd like to invite onstage

現在，我邀請Evan Grant

Evan Grant, who is one of last year's speakers,

去年的演講者上台

who's kindly agreed

他很樂意

to help me to demonstrate

幫忙示範

what we've been able to develop.

我們所設計的儀器

(Applause)

(鼓掌)

So the device that you see

你們所看到的儀器是

is a 14-channel, high-fidelity

有十四個頻道，高傳真的

EEG acquisition system.

腦電波訊號擷取系統

It doesn't require any scalp preparation,

不需要任何頭皮準備程序

no conductive gel or paste.

沒有導電的膠或漿糊

It only takes a few minutes to put on

戴上它，等訊號穩定

and for the signals to settle.

只要幾分鐘

It's also wireless,

而且是無線的

so it gives you the freedom to move around.

它讓你活動自如

And compared to the tens of thousands of dollars

比起那些幾萬美元的

for a traditional EEG system,

傳統腦電波系統

this headset only costs

這個頭戴式耳機

a few hundred dollars.

只要幾百美金

Now on to the detection algorithms.

現在來談談大腦感應演算法

So facial expressions --

好，面部表情--

as I mentioned before in emotional experiences --

如同之前講到的情緒經驗--

are actually designed to work out of the box

這套系統有令人意想不到的設計

with some sensitivity adjustments

只要做一些敏感度調整

available for personalization.

就可以運用於個人化的使用

But with the limited time we have available,

但因時間的關係

I'd like to show you the cognitive suite,

現在只示範認知的部份

which is the ability for you

這套系統能夠讓您

to basically move virtual objects with your mind.

只用意念移動虛擬物件

Now, Evan is new to this system,

Evan是第一次接觸這個系統

so what we have to do first

因此我們要先

is create a new profile for him.

建立一個新的檔案

He's obviously not Joanne -- so we'll "add user."

他當然不是Joanne， 所以要增加一個用戶

Evan. Okay.

Evan，好了!

So the first thing we need to do with the cognitive suite

首先要做的是

is to start with training

練習發出一個

a neutral signal.

中立的訊號

With neutral, there's nothing in particular

Evan不需要做

that Evan needs to do.

什麼特別的事

He just hangs out. He's relaxed.

就這樣放輕鬆

And the idea is to establish a baseline

重點是建立一個基準線

or normal state for his brain,

或是大腦的正常狀態

because every brain is different.

因為每個人的腦都不相同

It takes eight seconds to do this,

這大概需要八秒的時間

and now that that's done,

完成了

we can choose a movement-based action.

我們可以選擇一個有動作的活動

So Evan, choose something

Evan，你可選擇一個

that you can visualize clearly in your mind.

在你腦海中可以清楚看到的事情

Evan Grant: Let's do "pull."

讓我們做一個"拉"的動作

Tan Le: Okay, so let's choose "pull."

好，點選"拉"

So the idea here now

我們現在

is that Evan needs to

需要Evan想像

imagine the object coming forward

一件物品在螢幕上

into the screen,

往前移動

and there's a progress bar that will scroll across the screen

他這樣做的時候

while he's doing that.

螢幕上會出現一個測量棒

The first time, nothing will happen,

第一次沒有任何事情發生

because the system has no idea how he thinks about "pull."

因為系統還不知道他怎麼想像"拉"的動作

But maintain that thought

在這八秒中

for the entire duration of the eight seconds.

持續想著這個念頭

So: one, two, three, go.

一、二、三、開始

Okay.

好了

So once we accept this,

當我們按了接受

the cube is live.

這個方塊就活了起來

So let's see if Evan

讓我們看看Evan

can actually try and imagine pulling.

能否真的嘗試想像"拉"的動作

Ah, good job!

哇! 非常好!

(Applause)

(鼓掌)

That's really amazing.

真是令人驚訝！

(Applause)

(鼓掌)

So we have a little bit of time available,

我們還有一些時間

so I'm going to ask Evan

我要請Evan

to do a really difficult task.

做一些比較困難的動作

And this one is difficult

這個有點難

because it's all about being able to visualize something

因為要想像

that doesn't exist in our physical world.

在物質界裡不存在的事物

This is "disappear."

就是 "消失"

So what you want to do -- at least with movement-based actions,

就動作而言

we do that all the time, so you can visualize it.

因為經常做這些動作，所以能"看見"它

But with "disappear," there's really no analogies --

但"消失"沒有任何類似的動作

so Evan, what you want to do here

Evan, 現在請你

is to imagine the cube slowly fading out, okay.

想像這個方塊慢慢消失

Same sort of drill. So: one, two, three, go.

一樣的練習。 一、二、三、開始

Okay. Let's try that.

可以了，我們試試吧

Oh, my goodness. He's just too good.

我的天啊！他真的是非常厲害

Let's try that again.

再試一次

EG: Losing concentration.

（EG儀器:） 失去專注力

(Laughter)

(笑聲)

TL: But we can see that it actually works,

這套系統真的辦到了

even though you can only hold it

雖然只維持

for a little bit of time.

一段很短的時間

As I said, it's a very difficult process

我認為想像"消失"

to imagine this.

真的是非常困難

And the great thing about it is that

這個系統了不起的是

we've only given the software one instance

這套軟體只有一次機會

of how he thinks about "disappear."

知道Evan是怎麼想像"消失"的

As there is a machine learning algorithm in this --

而這部機器便學會了演算它

(Applause)

(鼓掌)

Thank you.

謝謝

Good job. Good job.

很棒！很棒！

(Applause)

(鼓掌)

Thank you, Evan, you're a wonderful, wonderful

謝謝，Evan你真的是這項科技

example of the technology.

最佳的展示人員

So, as you can see, before,

正如你們所見

there is a leveling system built into this software

這個軟體有一個水準測量系統

so that as Evan, or any user,

Evan或其他使用者

becomes more familiar with the system,

對這個系統越熟悉

they can continue to add more and more detections,

就能不斷地增加更多，更多的檢測項目

so that the system begins to differentiate

這個系統就能開始分辨

between different distinct thoughts.

不同的明顯想法

And once you've trained up the detections,

當你訓練做這些檢測項目

these thoughts can be assigned or mapped

這些念頭、想法就能指定或聯繫到

to any computing platform,

任何的電腦平台、

application or device.

應用程式或儀器上

So I'd like to show you a few examples,

讓我為你們展示幾個例子

because there are many possible applications

這個新界面有

for this new interface.

很多可運用的應用程式

In games and virtual worlds, for example,

例如在遊戲或虛擬世界

your facial expressions

你可以用臉部表情

can naturally and intuitively be used

自然、直覺地

to control an avatar or virtual character.

操控遊戲角色或虛擬人物

Obviously, you can experience the fantasy of magic

無庸置疑，你將會親身體驗幻想的魔力

and control the world with your mind.

和運用意念來控制世界

And also, colors, lighting,

顏色，燈光

sound and effects

聲音和音效

can dynamically respond to your emotional state

也可以不斷地變化來反映你的情緒狀態

to heighten the experience that you're having, in real time.

即時強化你的感受

And moving on to some applications

現在來看看應用程式

developed by developers and researchers around the world,

全世界的研發人員發明了

with robots and simple machines, for example --

不同的機械人和簡單的機器，例如

in this case, flying a toy helicopter

這個例子是操作玩具直昇機

simply by thinking "lift" with your mind.

只要用意念就可以讓它飛起來

The technology can also be applied

這項科技也可以應用在

to real world applications --

實際生活中

in this example, a smart home.

看看智能家居的例子

You know, from the user interface of the control system

從使用者界面控制系統

to opening curtains

來打開

or closing curtains.

或關上窗簾

And of course, also to the lighting --

當然電燈也可以

turning them on

開

or off.

或關

And finally,

最後

to real life-changing applications,

是應用在改善真實生活

such as being able to control an electric wheelchair.

例如能夠控制電動輪椅

In this example,

這個例子裡

facial expressions are mapped to the movement commands.

面部表情對應於移動方向的指令

Man: Now blink right to go right.

男聲: 現在眨右眼右轉

Now blink left to turn back left.

眨左眼左轉

Now smile to go straight.

微笑往前

TL: We really -- Thank you.

TL: 我們真的.... 多謝各位。

(Applause)

(鼓掌)

We are really only scratching the surface of what is possible today,

現今我們所做到的只是很小的一部分

and with the community's input,

有研發團隊的投入

and also with the involvement of developers

及全世界的研發和

and researchers from around the world,

研究人員的參與

we hope that you can help us to shape

我們希望這一項科技能夠

where the technology goes from here. Thank you so much.

從這裡一路順利發展。謝謝各位。