Why Don’t We Have Better Robots Yet? | Ken Goldberg | TED

183,935 views ・ 2024-03-28

TED


请双击下面的英文字幕来播放视频。

翻译人员: Yip Yan Yeung 校对人员: suya f.
00:04
I have a feeling most people in this room would like to have a robot at home.
0
4268
5739
我猜在座的大多数人 家里都有机器人。
00:10
It'd be nice to be able to do the chores and take care of things.
1
10041
3603
会做家务、 会处理事情真是太好了。
00:13
Where are these robots?
2
13677
1168
这些机器人在哪里?
00:14
What's taking so long?
3
14879
1601
为什么花了这么长的时间?
00:16
I mean, we have our tricorders,
4
16514
2636
我们有三轴飞行器, 还有卫星。
00:19
and we have satellites.
5
19183
2603
00:22
We have laser beams.
6
22253
2169
我们有激光束。
00:24
But where are the robots?
7
24755
1935
但是机器人在哪里?
00:26
(Laughter)
8
26690
1502
(笑声)
00:28
I mean, OK, wait, we do have some robots in our home,
9
28225
3170
好吧,等等,
我们家里确实有一些机器人,
00:31
but, not really doing anything that exciting, OK?
10
31395
4238
但是没在做那些酷炫的事,对吧?
00:35
(Laughter)
11
35666
1268
(笑声)
00:36
Now I've been doing research at UC Berkeley for 30 years
12
36967
4205
我已经在加州大学伯克利分校
和我的学生一起研究了 30 年的机器人,
00:41
with my students on robots,
13
41205
2703
00:43
and in the next 10 minutes,
14
43941
1435
在接下来的 10 分钟里,
00:45
I'm going to try to explain the gap between fiction and reality.
15
45409
4805
我将尝试解释 虚构与现实之间的差距。
00:50
Now we’ve seen images like this, right?
16
50514
2803
我们已经看到了 这样的影像,对吧?
00:53
These are real robots.
17
53350
1235
这些是真实的机器人。
00:54
They're pretty amazing.
18
54585
1201
它们太神奇了。
00:55
But those of us who work in the field,
19
55820
1868
但那些在该领域工作的人,
00:57
well, the reality is more like this.
20
57721
2036
好吧,现实更像这样。
00:59
(Laughter)
21
59757
2502
(笑声)
01:02
That's 99 out of 100 times, that's what happens.
22
62293
3203
100 次中的 99 次是这样, 事实就是这样。
01:05
And in the field, there's something that explains this
23
65529
2836
在这个领域,
01:08
that we call Moravec's paradox.
24
68399
2302
“莫拉维克悖论” 可以解释这个现象。
01:10
And that is, what's easy for robots,
25
70734
2069
对机器人来说很容易的任务,
01:12
like being able to pick up a large object,
26
72837
3570
比如捡起一个大型物体,
01:16
large, heavy object,
27
76440
1268
大而重的物体,
01:17
is hard for humans.
28
77708
2202
对人类来说是很难的。
01:20
But what's easy for humans,
29
80377
2369
但是对人类很容易的任务,
01:22
like being able to pick up some blocks and stack them,
30
82780
3804
比如捡起一些方块, 把它们叠起来,
01:26
well, it turns out that is very hard for robots.
31
86584
3970
对机器人来说非常困难。
01:31
And this is a persistent problem.
32
91388
2236
这是一个持续存在的问题。
01:33
So the ability to grasp arbitrary objects is a grand challenge for my field.
33
93657
6307
抓住任意物体的能力 在我这个领域是一项艰巨的挑战。
01:40
Now by the way, I was a very klutzy kid.
34
100531
4204
顺便说一句, 我是一个笨手笨脚的孩子。
01:44
(Laughter)
35
104735
1301
(笑声)
01:46
I would drop things.
36
106070
1168
我会拿不住东西。
01:47
Any time someone would throw me a ball, I would drop it.
37
107271
2636
只要有人丢给我一个球, 我都接不住。
01:49
I was the last kid to get picked on a basketball team.
38
109940
2703
我是最后一个 被选入篮球队的孩子。
01:52
I'm still pretty klutzy, actually,
39
112676
1635
我其实还是很笨拙,
01:54
but I have spent my entire career studying how to make robots less clumsy.
40
114345
5472
但我整个职业生涯都在研究 如何让机器人不那么笨拙。
02:00
Now let's start with the hardware.
41
120484
1935
让我们从硬件开始。
02:02
So the hands.
42
122453
2069
双手。
02:04
Now this is a robot hand, a particular type of hand.
43
124555
3337
这是一只机器人的手, 一种特殊的手。
02:07
It's a lot like our hand.
44
127925
1235
这很像我们的手。
02:09
And it has a lot of motors, a lot of tendons
45
129193
3470
你可以看到它有很多马达、 很多肌腱和电缆。
02:12
and cables as you can see.
46
132696
1535
02:14
So it's unfortunately not very reliable.
47
134265
2602
不幸的是,它不是很可靠。
02:16
It's also very heavy and very expensive.
48
136901
2435
它还很重,而且非常昂贵。
02:19
So I'm in favor of very simple hands, like this.
49
139637
3803
我更喜欢更简单的手,就像这个。
02:23
So this has just two fingers.
50
143440
2403
它只有两根手指。
02:25
It's known as a parallel jaw gripper.
51
145876
1769
它被称为“平行夹爪”。
02:28
So it's very simple.
52
148012
1535
它很简单。
02:29
It's lightweight and reliable and it's very inexpensive.
53
149580
4438
它轻巧可靠,而且非常便宜。
02:34
And if you're doubting that simple hands can be effective,
54
154652
3970
如果你在怀疑 这简单的手有多有效,
02:38
look at this video where you can see that two very simple grippers,
55
158622
4371
看看这个视频,你可以看到 两个非常简单的夹爪,
02:43
these are being operated, by the way,
56
163027
1935
顺便提一句,它们是由人 像控制木偶那样操控的。
02:44
by humans who are controlling the grippers like a puppet.
57
164995
2903
02:47
But very simple grippers are capable of doing very complex things.
58
167898
3504
但是非常简单的夹爪 能够做非常复杂的事情。
02:51
Now actually in industry,
59
171435
1468
其实在行业中,
02:52
there’s even a simpler robot gripper, and that’s the suction cup.
60
172937
3703
还有一种更简单的机械夹爪, 那就是吸盘。
02:56
And that only makes a single point of contact.
61
176674
2469
只需要一个接触点。
02:59
So again, simplicity is very helpful in our field.
62
179176
3037
同样,“简单” 在我们的领域中非常有用。
03:02
Now let's talk about the software.
63
182646
2169
我们来聊聊软件。
03:04
This is where it gets really, really difficult
64
184815
3203
这就是非常困难的地方,
03:08
because of a fundamental issue, which is uncertainty.
65
188018
4004
源于一个根本性的问题, 那就是“不确定性”。
03:12
There's uncertainty in the control.
66
192356
2202
控制中存在不确定性。
03:14
There’s uncertainty in the perception.
67
194592
2135
感知中存在不确定性。
03:16
And there’s uncertainty in the physics.
68
196760
2169
物理上存在不确定性。
03:19
Now what do I mean by the control?
69
199463
1668
我说的“控制”是什么意思?
03:21
Well if you look at a robot’s gripper trying to do something,
70
201165
3670
如果你看看机器人的抓手尝试做某事,
03:24
there's a lot of uncertainty in the cables and the mechanisms
71
204868
3904
就会发现电缆和机械装置 存在着很大的不确定性,
03:28
that cause very small errors.
72
208806
2102
而这些不确定性会导致非常小的错误。
03:30
And these can accumulate and make it very difficult to manipulate things.
73
210908
4104
而且它们会累积起来, 使操纵物体变得非常困难。
03:36
Now in terms of the sensors, yes,
74
216046
2002
就传感器而言,没错,
03:38
robots have very high-resolution cameras just like we do,
75
218082
3803
机器人像我们一样, 拥有非常高分辨率的摄像头,
03:41
and that allows them to take images of scenes in traffic
76
221919
3904
使它们能够拍摄交通、
03:45
or in a retirement center,
77
225856
1668
退休中心、
03:47
or in a warehouse or in an operating room.
78
227558
3136
仓库或手术室中的场景图像。
03:50
But these don't give you the three-dimensional structure
79
230694
2670
但是这些并不能给你 现况的三维结构。
03:53
of what's going on.
80
233364
1334
03:54
So recently, there was a new development called LIDAR,
81
234999
3503
最近,有一项名为 “激光雷达”(LIDAR)的新技术,
03:58
and this is a new class of cameras that use light beams to build up
82
238535
4505
是一种新型摄像机,使用光束
04:03
a three-dimensional model of the environment.
83
243073
2436
建立环境的三维模型。
04:06
And these are fairly effective.
84
246277
1768
它们相当有效。
04:08
They really were a breakthrough in our field, but they're not perfect.
85
248078
4171
它们确实是我们领域的一项突破, 但它们并不完美。
04:12
So if the objects have anything that's shiny or transparent,
86
252283
5005
如果物体有任何闪亮或透明的东西,
04:17
well, then the light acts in unpredictable ways,
87
257288
2435
那么,光线就会 出现无法预测的情况,
04:19
and it ends up with noise and holes in the images.
88
259757
2335
最终图像中会产生噪点和孔洞。
04:22
So these aren't really the silver bullet.
89
262126
2002
因此,它们并不是真正的万灵药。
04:24
And there’s one other form of sensor out there now called a “tactile sensor.”
90
264595
4171
现在还有另一种形式的传感器, 叫做“触觉传感器”。
04:28
And these are very interesting.
91
268766
1501
它们非常有趣。
04:30
They use cameras to actually image the surfaces
92
270301
3203
它们用摄像头对表面摄影,
04:33
as a robot would make contact,
93
273504
2135
就像机器人接触表面那样,
04:35
but these are still in their infancy.
94
275673
1935
但它们仍处于起步阶段。
04:38
Now the last issue is the physics.
95
278242
2702
最后一个问题是物理。
04:40
And let me illustrate for you by showing you,
96
280978
3203
我来展示一下,
04:44
we take a bottle on a table
97
284214
1769
我们把一个瓶子放在桌上,
04:45
and we just push it,
98
285983
1168
推它,
04:47
and the robot's pushing it in exactly the same way each time.
99
287184
3403
机器人每次的推动方式完全一样。
04:50
But you can see that the bottle ends up in a very different place each time.
100
290621
4471
但是你可以看到,瓶子每次都会 出现在一个截然不同的地方。
04:55
And why is that?
101
295125
1168
为什么?
04:56
Well it’s because it depends on the microscopic surface topography
102
296327
4871
这是因为这取决于在瓶子滑动时,
它下方出现的微观表面地形。
05:01
underneath the bottle as it slid.
103
301198
2603
05:03
For example, if you put a grain of sand under there,
104
303834
2469
比如,如果你在下面放了一粒沙子,
05:06
it would react very differently than if there weren't a grain of sand.
105
306337
3536
相较没有那粒沙子, 反应会截然不同。
05:09
And we can't see if there's a grain of sand because it's under the bottle.
106
309873
3804
我们看不出是否有沙子, 因为它在瓶子下面。
05:14
It turns out that we can predict the motion of an asteroid
107
314311
3937
事实证明,我们可以预测 一百万英里以外的小行星的运动,
05:18
a million miles away,
108
318248
2636
05:20
far better than we can predict the motion of an object
109
320918
3170
效果远比我们预测一个物体
05:24
as it's being grasped by a robot.
110
324121
2002
被机器人抓住时的运动要好得多。
05:27
Now let me give you an example.
111
327391
2169
我来举个例子。
05:29
Put yourself here into the position of being a robot.
112
329593
3537
把自己置于机器人的角色。
05:33
You're trying to clear the table
113
333731
1534
你想清理桌子,
05:35
and your sensors are noisy and imprecise.
114
335265
2570
但你的传感器噪声很大, 而且不精确。
05:37
Your actuators, your cables and motors are uncertain,
115
337835
3436
你的驱动器、电缆和电机不稳定,
05:41
so you can't fully control your own gripper.
116
341305
2369
所以你无法完全控制自己的夹爪。
05:43
And there's uncertainty in the physics,
117
343707
2102
物理上也存在不确定性,
05:45
so you really don't know what's going to happen.
118
345809
2303
所以你真的不知道会发生什么。
05:48
So it's not surprising that robots are still very clumsy.
119
348145
3604
因此,机器人仍然非常笨拙 也就不足为奇了。
05:52
Now there's one sweet spot for robots, and that has to do with e-commerce.
120
352883
4805
机器人有一个最佳用处, 与电子商务有关。
05:57
And this has been growing, it's a huge trend.
121
357721
2102
这种趋势一直在增长, 是一个巨大的趋势。
05:59
And during the pandemic, it really jumped up.
122
359857
3036
在疫情期间,它突飞猛进。
06:02
I think most of us can relate to that.
123
362926
2336
我认为我们很多人都有所体会。
06:05
We started ordering things like never before,
124
365829
2570
我们比以往都更常订购商品,
06:08
and this trend is continuing.
125
368432
1735
而且这种趋势仍在继续。
06:10
And the challenge is to meet the demand,
126
370200
2770
挑战在于满足需求,
06:13
we have to be able to get all these packages delivered in a timely manner.
127
373003
4972
我们必须及时交付所有包裹。
06:18
And the challenge is that every package is different,
128
378575
2503
挑战在于每个包裹都不一样,
06:21
every order is different.
129
381078
1234
每个订单都不一样。
06:22
So you might order some some nail polish and an electric screwdriver.
130
382346
5939
你可以订购一些指甲油 和一把电动螺丝刀。
06:28
And those two objects are going to be
131
388652
2770
而这两个商品将位于
06:31
somewhere inside one of these giant warehouses.
132
391455
3036
一个巨型仓库内的某个位置。
06:34
And what needs to be done is someone has to go in,
133
394925
2336
得有人进去,
06:37
find the nail polish and then go and find the screwdriver,
134
397294
2769
找到指甲油、找到螺丝刀,
06:40
bring them together, put them into a box and deliver them to you.
135
400063
3204
把它们放在一起、
放到一个盒子里、 交到你手上。
06:43
So this is extremely difficult, and it requires grasping.
136
403300
3137
这是极其困难的,需要抓取的动作。
06:46
So today, this is almost entirely done with humans.
137
406437
3136
眼下几乎完全由人类完成。
06:49
And the humans don't like doing this work,
138
409606
2002
而且人类不喜欢做这项工作,
06:51
there's a huge amount of turnover.
139
411642
1768
流失率很大。
06:53
So it's a challenge.
140
413410
1201
因此,这是一个挑战。
06:54
And people have tried to put robots
141
414611
2937
人们试图将机器人 放入仓库来完成这项工作。
06:57
into warehouses to do this work.
142
417581
3070
07:01
(Laughter)
143
421018
6339
(笑声)
07:08
It hasn't turned out all that well.
144
428192
3937
结果没那么好。
07:12
But my students and I, about five years ago,
145
432563
3570
但是,大约五年前,我和我的学生
07:16
we came up with a method, using advances in AI and deep learning,
146
436133
4071
想出了一种方法, 利用 AI 和深度学习的进步,
07:20
to have a robot essentially train itself to be able to grasp objects.
147
440204
3903
让机器人进行自我训练, 使其能够抓住物体。
07:24
And the idea was that the robot would do this in simulation.
148
444441
2836
我们想的是机器人 通过模拟进行这一过程。
07:27
It was almost as if the robot were dreaming about how to grasp things
149
447277
3270
就像是机器人想象如何抓住物体,
07:30
and learning how to grasp them reliably.
150
450581
1935
并学习如何稳定地抓住它们。
07:32
And here's the result.
151
452816
1168
以下是结果。
07:34
This is a system called Dex-net
152
454017
1869
这是一个名为 DEX-net 的系统,
07:35
that is able to reliably pick up objects
153
455919
3270
它能够稳定地拾取
07:39
that we put into these bins in front of the robot.
154
459223
2502
我们放在机器人前面的桶里的物体,
07:41
These are objects it's never been trained on,
155
461758
2670
这些是它从未接受过训练的物体,
07:44
and it's able to pick these objects up
156
464461
1835
它能够一遍又一遍地 捡起这些物体
07:46
and reliably clear these bins over and over again.
157
466330
2869
并稳定地清空这几个桶。
07:49
So we were very excited about this result.
158
469666
2570
我们对这个结果感到非常兴奋。
07:52
And the students and I went out to form a company,
159
472269
3003
我和学生们 出去创立了一家公司,
07:55
and we now have a company called Ambi Robotics.
160
475305
3103
现在我们有一家 名为 Ambi Robotics 的公司。
07:58
And what we do is make machines that use the algorithms,
161
478742
4104
我们让机器通过算法,
08:02
the software we developed at Berkeley,
162
482880
2268
即我们在伯克利开发的软件,
08:05
to pick up packages.
163
485148
1936
拾取包裹。
08:07
And this is for e-commerce.
164
487117
1902
这是为电子商务开发的。
08:09
The packages arrive in large bins, all different shapes and sizes,
165
489052
3504
这些包裹装在形状和大小 各不相同的大箱子里,
08:12
and they have to be picked up,
166
492589
1468
必须根据邮编将它们拾起、 扫描,然后放入较小的箱子里。
08:14
scanned and then put into smaller bins depending on their zip code.
167
494057
3337
08:18
We now have 80 of these machines operating across the United States,
168
498061
4205
我们在美国各地 有 80 台此类机器在运行,
08:22
sorting over a million packages a week.
169
502299
2836
每周分拣超过一百万个包裹。
08:26
Now that’s some progress,
170
506169
3204
已经取得了一些进展,
08:29
but it's not exactly the home robot that we've all been waiting for.
171
509406
3837
但这并不完全是 我们一直在期待的家用机器人。
08:33
So I want to give you a little bit of an idea
172
513677
2536
因此,我想向你们介绍一下 我们正在进行的一些新研究,
08:36
of some the new research that we're doing
173
516246
2102
08:38
to try to be able to have robots more capable in homes.
174
518348
3070
目的是让机器人 在家中发挥更大的能力。
08:41
And one particular challenge is being able to manipulate deformable objects,
175
521752
4171
一个特别的挑战是 操纵可变形的物体,
08:45
like strings in one dimension,
176
525956
2102
例如一维的绳索、
08:48
two-dimensional sheets and three dimensions,
177
528058
3470
二维的纸张和三维的物体,
08:51
like fruits and vegetables.
178
531562
2202
例如水果和蔬菜。
08:53
So we've been working on a project to untangle knots.
179
533764
3804
我们在做一个解开绳结的项目。
08:57
And what we do is we take a cable and we put that in front of the robot.
180
537901
4104
我们要做的是把一根电缆 放在机器人前面。
09:02
It has to use a camera to look down, analyze the cable,
181
542039
2936
它得用摄像头向下看、 分析电缆、
09:04
figure out where to grasp it
182
544975
1368
想出抓住什么位置、
09:06
and how to pull it apart to be able to untangle it.
183
546343
2836
如何把它抻开才能解开。
09:09
And this is a very hard problem,
184
549580
1668
这是一个非常棘手的问题,
09:11
because the cable is much longer than the reach of the robot.
185
551248
2903
因为电缆比机器人的 触及范围要长得多。
09:14
So it has to go through and manipulate, manage the slack as it's working.
186
554184
4438
它必须仔细检查和操纵, 在操作过程中留意松弛的部分。
09:18
And I would say this is doing pretty well.
187
558655
2436
我觉得它做得还挺不错。
09:21
It's gotten up to about 80 percent success
188
561124
2269
可以达到约 80% 的成功率,
09:23
when we give it a tangled cable at being able to untangle it.
189
563427
3203
让它解开一条打结的电缆。
09:27
The other one is something I think we also all are waiting for:
190
567731
3070
我觉得我们还在等待另一个机器人:
09:30
robot to fold the laundry.
191
570834
2269
叠衣服的机器人。
09:33
Now roboticists have actually been looking at this for a long time,
192
573136
4071
机器人专家其实已经研究了很长时间,
09:37
and there was some research that was done on this.
193
577240
3037
并且已经对此进行了一些研究。
09:40
But the problem is that it's very, very slow.
194
580277
2869
但问题在于它非常非常慢。
09:43
So this was about three to six folds per hour.
195
583180
5572
一个小时叠 3 到 6 次。
09:48
(Laughter)
196
588785
1902
(笑声)
09:50
So we decided to to revisit this problem
197
590721
3403
我们决定重新审视这个问题,
09:54
and try to have a robot work very fast.
198
594124
1902
试着让机器人快速工作。
09:56
So one of the things we did was try to think
199
596059
2069
我们做的一件事是试着
09:58
about a two-armed robot that could fling the fabric
200
598128
2402
让一台双臂机器人 像我们叠衣服的时候抖一抖衣物,
10:00
the way we do when we're folding,
201
600530
1602
10:02
and then we also used friction in this case to drag the fabric
202
602132
2903
我们还利用摩擦力拖动衣物,
10:05
to smooth out some wrinkles.
203
605035
1602
扯平褶子。
10:06
And then we borrowed a trick which is known as the two-second fold.
204
606670
4838
我们还参考了 “两秒叠衣服”的技巧。
10:11
You might have heard of this.
205
611541
1402
你可能听说过这个。
10:12
It's amazing because the robot is doing exactly the same thing
206
612976
3270
太神奇了, 因为机器人就在做这件事情,
10:16
and it's a little bit longer, but that's real time,
207
616279
2536
虽然时间长了点, 但这个是原倍速的。
10:18
it's not sped up.
208
618849
1168
没有开倍速。
10:20
So we're making some progress there.
209
620017
1968
我们已经取得了一些进展。
10:23
And the last example is bagging.
210
623120
1601
最后一个例子是装袋。
10:24
So you all encounter this all the time.
211
624755
2168
所有人都总是会遇到这种情况。
10:26
You go to a corner store, and you have to put something in a bag.
212
626957
3070
你去街角的商店, 你要在袋子里放点东西。
10:30
Now it's easy, again, for humans,
213
630027
1601
对于人类来说,这很容易,
10:31
but it's actually very, very tricky for robots
214
631628
3370
但其实对机器人来说 非常非常棘手,
10:35
because for humans, you know how to take the bag
215
635032
2268
因为对于人类来说, 你知道如何拿起袋子、
10:37
and how to manipulate it.
216
637300
1235
如何操纵它。
10:38
But robots, the bag can arrive in many different configurations.
217
638568
3104
但是对机器人来说, 袋子会以不同的形态出现。
10:41
It’s very hard to tell what’s going on
218
641705
2836
很难分辨发生了什么,
10:44
and for the robot to figure out how to open up that bag.
219
644541
2769
机器人也很难弄清楚 如何打开那个袋子。
10:47
So what we did was we had the robot train itself.
220
647310
4705
于是我们让机器人自我训练。
10:52
We painted one of these bags with fluorescent paint,
221
652015
2469
我们在袋子上涂了荧光涂料,
10:54
and we had fluorescent lights that would turn on and off,
222
654518
2702
配上可以开关的荧光灯,
10:57
and the robot would essentially teach itself how to manipulate these bags.
223
657220
4438
机器人自学如何操纵这些袋子。
11:01
And so we’ve gotten it now up to the point
224
661692
2035
我们现在已经能让它
11:03
where we're able to solve this problem about half the time.
225
663727
3770
解决这个问题的速度提速一倍。
11:07
So it works,
226
667497
1268
有效,但我得说我们还没完全达成。
11:08
but I'm saying, we're still not quite there yet.
227
668765
3737
11:12
So I want to come back to Moravec's paradox.
228
672502
2303
我想回过头来谈谈莫拉维克悖论。
11:14
What's easy for robots is hard for humans.
229
674838
2236
对机器人来说容易的事情 对人类来说却很难。
11:17
And what's easy for us is still hard for robots.
230
677107
4871
对我们来说容易的事情 对机器人来说很难。
11:22
We have incredible capabilities.
231
682412
2369
我们拥有卓越的能力。
11:24
We're very good at manipulation.
232
684781
1669
我们非常擅长操纵。
11:26
(Laughter)
233
686483
1935
(笑声)
11:28
But robots still are not.
234
688418
1769
但是机器人仍然不是。
11:31
I want to say, I understand.
235
691421
2236
我想说,我明白。
11:33
It’s been 60 years,
236
693990
2002
已经过去了 60 年,
11:35
and we're still waiting for the robots that the Jetsons had.
237
695992
3838
我们仍在等待 杰森一家拥有的机器人。
11:40
Why is this difficult?
238
700497
1368
为什么这很难?
11:41
We need robots because we want them to be able to do tasks that we can't do
239
701898
6473
我们之所以需要机器人,是因为 我们希望它们完成我们无法完成
11:48
or we don't really want to do.
240
708405
1835
或不太想做的任务。
11:50
But I want you to keep in mind that these robots, they're coming.
241
710841
3837
但我希望你知道, 这些机器人,它们即将到来。
11:54
Just be patient.
242
714711
1368
耐心点。
11:56
Because we want the robots,
243
716379
2103
因为我们想要机器人,
11:58
but robots also need us
244
718515
2169
但机器人也需要我们
12:00
to do the many things that robots still can't do.
245
720717
4405
去做许多机器人仍然做不到的事情。
12:06
Thank you.
246
726156
1234
谢谢。
12:07
(Applause)
247
727390
2436
(掌声)
关于本网站

这个网站将向你介绍对学习英语有用的YouTube视频。你将看到来自世界各地的一流教师教授的英语课程。双击每个视频页面上显示的英文字幕,即可从那里播放视频。字幕会随着视频的播放而同步滚动。如果你有任何意见或要求,请使用此联系表与我们联系。

https://forms.gle/WvT1wiN1qDtmnspy7