TEDx] スマートマシン、未来を変える：コリン・アングルがTEDxTaipei 2013に登場 (【TEDx】智慧機器，轉變未來：Colin Angle at TEDxTaipei 2013)

字幕表動画を再生する

1
00:00:17,043 --> 00:00:19,756
My name is Colin Angle and
2
00:00:19,756 --> 00:00:23,709
I founded my company 'iRobot' 23 years ago
3
00:00:23,709 --> 00:00:28,793
with a single focus to make robots practical.
4
00:00:28,793 --> 00:00:34,840
In 1962, there was an American cartoon,
5
00:00:34,840 --> 00:00:38,626
called The Jetsons with Rosie the robot.
6
00:00:38,626 --> 00:00:41,681
And when the world saw Rosie,
7
00:00:41,681 --> 00:00:46,517
we all knew what the next big thing was going to be.
8
00:00:46,517 --> 00:00:50,792
Unfortunately 40 years passed,
9
00:00:50,792 --> 00:00:53,542
and it didn't happen.
10
00:00:53,542 --> 00:00:56,542
What I'd like to spend a few minutes with you today talking about,
11
00:00:56,542 --> 00:01:03,543
is the acceleration and coming of age of the robot industry.
12
00:01:03,543 --> 00:01:05,668
We are starting to see around the world
13
00:01:05,668 --> 00:01:11,442
applications where robots are actually making a difference.
14
00:01:11,442 --> 00:01:15,347
And we are actually seeing exciting new businesses
15
00:01:15,347 --> 00:01:20,619
spring up and change how we think about the world and how
16
00:01:20,619 --> 00:01:25,042
we think about machines.
17
00:01:25,042 --> 00:01:33,377
The important realization was that robots do not look like this.
18
00:01:33,377 --> 00:01:37,767
Instead, if you want to make a business, if you want robots in our world
19
00:01:37,767 --> 00:01:41,292
we need to think about what are the challenges
20
00:01:41,292 --> 00:01:45,776
that face people on a daily basis,
21
00:01:45,776 --> 00:01:49,279
and to think about how
22
00:01:49,279 --> 00:01:52,182
we can create intelligent machines
23
00:01:52,182 --> 00:01:57,792
that can deliver more value then they cost to produce.
24
00:01:57,792 --> 00:01:59,376
If we can do that,
25
00:01:59,376 --> 00:02:02,662
we have created a great business.
26
00:02:02,662 --> 00:02:07,958
If we make something that is an amazing and exciting demonstration,
27
00:02:07,958 --> 00:02:11,459
we have made an exciting demonstration that had
28
00:02:11,459 --> 00:02:15,876
very little impact on the world.
29
00:02:15,876 --> 00:02:20,011
I believe that for every successful robot product
30
00:02:20,011 --> 00:02:25,295
there are many thousands of demonstration of robots.
31
00:02:25,295 --> 00:02:28,961
To give you an example of a success,
32
00:02:28,961 --> 00:02:35,047
where by abandoning traditional notions of what a robot should be
33
00:02:35,047 --> 00:02:38,548
we have created something of real impact,
34
00:02:38,548 --> 00:02:41,766
we can look at vacuuming robots.
35
00:02:41,766 --> 00:02:43,797
So this is the 'Roomba'.
36
00:02:43,797 --> 00:02:48,572
This is my robot. We've sold over 10 million of these robots.
37
00:02:48,572 --> 00:02:52,341
And I think people would find it surprising if I told you
38
00:02:52,341 --> 00:02:56,714
that 15 percent of all of the money spent in the world
39
00:02:56,714 --> 00:03:01,547
on vacuuming cleaners today is now spent on robots.
40
00:03:01,547 --> 00:03:03,048
And in fact,
41
00:03:03,048 --> 00:03:07,048
in a few countries
42
00:03:07,048 --> 00:03:10,994
the number one selling vacuuming the entire country is now
43
00:03:10,994 --> 00:03:12,846
a robot.
44
00:03:12,846 --> 00:03:17,680
This is an example of what can happen
45
00:03:17,680 --> 00:03:20,679
when you actually get it right.
46
00:03:20,679 --> 00:03:22,680
You actually say, well, people don't want to vacuum,
47
00:03:22,680 --> 00:03:25,512
people don't have time to vacuum.
48
00:03:25,512 --> 00:03:27,846
So how do we solve that?
49
00:03:27,846 --> 00:03:29,982
Another example
50
00:03:29,982 --> 00:03:34,288
of where robots are creating value
51
00:03:34,288 --> 00:03:38,723
is in very dangerous situations.
52
00:03:38,723 --> 00:03:41,292
In 2002,
53
00:03:41,292 --> 00:03:45,496
we… there was a great challenge in Afghanistan. The United States
54
00:03:45,496 --> 00:03:47,465
sent soldiers to Afghanistan.
55
00:03:47,465 --> 00:03:49,802
And we were watching television and they told us how
56
00:03:49,802 --> 00:03:52,262
the soldiers would have ropes
57
00:03:52,262 --> 00:03:56,374
tied around their waists to go into the caves in Afghanistan.
58
00:03:56,374 --> 00:03:58,512
So if they were hurt or shot,
59
00:03:58,512 --> 00:04:01,430
you could drag the solider out
60
00:04:01,430 --> 00:04:03,380
using that rope.
61
00:04:03,380 --> 00:04:06,720
And we said that was a crazy, crazy thing.
62
00:04:06,720 --> 00:04:09,346
Why has not technology solved this problem?
63
00:04:09,346 --> 00:04:12,122
Why are people doing that instead of a robot?
64
00:04:12,122 --> 00:04:14,826
So, we had a project to make a robot,
65
00:04:14,826 --> 00:04:16,793
we sent it over to Afghanistan,
66
00:04:16,793 --> 00:04:22,099
we sent it into caves. We found some very, very scary things.
67
00:04:22,099 --> 00:04:26,034
This is all automatic weapon, ammunition, and explosives
68
00:04:26,034 --> 00:04:29,394
founded in a cave by the robot.
69
00:04:29,394 --> 00:04:31,441
And we made a difference.
70
00:04:31,441 --> 00:04:32,876
Perhaps,
71
00:04:32,876 --> 00:04:38,977
the most easy to understand and graphical way of the difference
72
00:04:38,977 --> 00:04:42,227
I will demonstrate with two short videos.
73
00:04:42,227 --> 00:04:46,309
Both happened in the streets of Iraq.
74
00:04:46,309 --> 00:04:52,096
The first demonstrated here where you have a humvee,
75
00:04:52,096 --> 00:04:53,478
(can we have the sound up?)
76
00:04:53,478 --> 00:04:58,133
Um, you can see pushing a bomb off the street.
77
00:04:58,133 --> 00:05:02,726
78
00:05:02,726 --> 00:05:05,776
Now, contrast that video,
79
00:05:05,776 --> 00:05:07,180
with this video,
80
00:05:07,180 --> 00:05:10,310
of a robot doing the same sort of thing.
81
00:05:10,310 --> 00:05:13,017
(again the volume there)
82
00:05:13,017 --> 00:05:17,287
So you can see by that orange can is the robot,
83
00:05:17,287 --> 00:05:35,006
84
00:05:35,006 --> 00:05:38,009
So, in the first video you heard screams
85
00:05:38,009 --> 00:05:41,545
the driver of that vehicle was not killed
86
00:05:41,545 --> 00:05:45,084
um survived. But here we have laughter and guys think
87
00:05:45,084 --> 00:05:48,786
'Dude, Dude! I caught that on video!'
88
00:05:48,786 --> 00:05:53,894
So the robot was in fact destroyed but other than that
89
00:05:53,894 --> 00:05:57,127
we had resolved the situation.
90
00:05:57,127 --> 00:06:00,859
So again, robots for going into very scary and
91
00:06:00,859 --> 00:06:05,502
dangerous situations enable to change the equation
92
00:06:05,502 --> 00:06:10,943
and protect the lives of people trying to protect us.
93
00:06:10,943 --> 00:06:15,443
So this technology, this idea
94
00:06:15,443 --> 00:06:20,716
that was represented by those robots that you just saw
95
00:06:20,716 --> 00:06:25,423
is the beginning of a very exciting idea.
96
00:06:25,423 --> 00:06:31,359
We can create robots that can act as surrogates
97
00:06:31,359 --> 00:06:34,598
for our physical self
98
00:06:34,598 --> 00:06:39,775
and change the meaning of what being someplace means.
99
00:06:39,775 --> 00:06:45,193
So the next industry which is
100
00:06:45,193 --> 00:06:49,110
going to be changed by robots is health care.
101
00:06:49,110 --> 00:06:53,083
We're seeing the world where the sophistication
102
00:06:53,083 --> 00:06:56,587
of health care is increasing every single day.
103
00:06:56,587 --> 00:07:00,925
Treatments are more and more sophisticated.
104
00:07:00,925 --> 00:07:04,192
But it requires doctors
105
00:07:04,192 --> 00:07:07,566
with an increasingly sophisticated understanding
106
00:07:07,566 --> 00:07:12,239
of a narrow area of medicine to make sure that
107
00:07:12,239 --> 00:07:15,472
the patient gets the right treatment.
108
00:07:15,472 --> 00:07:18,142
This is causing a new challenge
109
00:07:18,142 --> 00:07:20,544
in medicine, trying to match up
110
00:07:20,544 --> 00:07:28,387
the specialist with the patient that needs his or her expertise.
111
00:07:28,387 --> 00:07:32,558
So what a great opportunity to create a robot, not to defuse
112
00:07:32,558 --> 00:07:36,627
bombs, but to diagnose patients at a distance.
113
00:07:36,627 --> 00:07:39,128
So this robot that you see
114
00:07:39,128 --> 00:07:43,800
behind me, is deployed. There's about 50 of these robots
115
00:07:43,800 --> 00:07:46,636
in action today in many different hospitals and
116
00:07:46,636 --> 00:07:50,974
the United States and Mexico
117
00:07:50,974 --> 00:07:53,944
and growing into an increasing number of hospitals.
118
00:07:53,944 --> 00:07:56,614
And the way they work
119
00:07:56,614 --> 00:08:00,500
is a big city hospital which has these specialists
120
00:08:00,500 --> 00:08:05,655
actually sends the robots to rural hospitals.
121
00:08:05,655 --> 00:08:08,357
Now that rural hospital can
122
00:08:08,357 --> 00:08:14,032
see a patient. Say right now, the most common uses is for stroke.
123
00:08:14,032 --> 00:08:17,136
A patient goes to that small hospital.
124
00:08:17,136 --> 00:08:21,572
Instead of the doctor walking in to see the patient,
125
00:08:21,572 --> 00:08:24,187
a robot drives in,
126
00:08:24,187 --> 00:08:30,648
performs the diagnosis. The right drug, or the right treatment
127
00:08:30,648 --> 00:08:35,105
is applied to help that patient be stable.
128
00:08:35,105 --> 00:08:38,156
And then if the patient can stay in that rural hospital,
129
00:08:38,156 --> 00:08:40,440
he or she does. If he needs additional treatment,
130
00:08:40,440 --> 00:08:47,164
he can now be safely moved to the larger hospital.
131
00:08:47,164 --> 00:08:54,038
We have seen dramatic increases in the outcomes
132
00:08:54,038 --> 00:09:00,880
of patients in these rural hospitals using this robot technology.
133
00:09:00,880 --> 00:09:04,821
And we should expect to see robots play a larger role
134
00:09:04,821 --> 00:09:06,238
in providing health care,
135
00:09:06,238 --> 00:09:08,218
because it will allow us to leverage fully
136
00:09:08,218 --> 00:09:12,572
the education of the doctors
137
00:09:12,572 --> 00:09:21,405
around the world to be put to use in the best possible way.
138
00:09:21,405 --> 00:09:24,367
How about something a little closer to home?
139
00:09:24,367 --> 00:09:28,655
Why do we all have to travel
140
00:09:28,655 --> 00:09:31,655
if we want to go somewhere?
141
00:09:31,655 --> 00:09:35,739
This is a new idea where
142
00:09:35,739 --> 00:09:39,549
I could be giving this talk to you today
143
00:09:39,549 --> 00:09:41,718
as a robot
144
00:09:41,718 --> 00:09:46,737
and able to do a different talk
145
00:09:46,737 --> 00:09:49,028
on the other side of the world tomorrow
146
00:09:49,028 --> 00:09:52,696
using a robot. This idea of making the world smaller
147
00:09:52,696 --> 00:09:57,822
not just for doctors, but for anyone,
148
00:09:57,822 --> 00:09:59,736
is a very exciting idea.
149
00:09:59,736 --> 00:10:03,905
So we've created a robot. And I have given
150
00:10:03,905 --> 00:10:06,842
a number of talks using this robot.
151
00:10:06,842 --> 00:10:08,488
And it's very exciting. I can stand up,
152
00:10:08,488 --> 00:10:12,048
I can give my talk, I can go answer questions afterwards.
153
00:10:12,048 --> 00:10:15,219
When my time is up, I can drive out of this hall
154
00:10:15,219 --> 00:10:19,057
and meet other people and answer their questions
155
00:10:19,057 --> 00:10:24,162
and behave exactly like I would if I was here in person.
156
00:10:24,162 --> 00:10:28,030
And as a result, I make the world a little smaller,
157
00:10:28,030 --> 00:10:32,903
and I increase our ability to communicate with each other.
158
00:10:32,903 --> 00:10:36,957
So the robot industry is changing.
159
00:10:36,957 --> 00:10:40,745
The robot industry is changing other industries
160
00:10:40,745 --> 00:10:43,791
but it's also being changed itself,
161
00:10:43,791 --> 00:10:44,792
which is very exciting.
162
00:10:44,792 --> 00:10:47,625
In 1991 I founded iRobot
163
00:10:47,625 --> 00:10:50,706
if I wanted my robot to be wirelessly connected
164
00:10:50,706 --> 00:10:54,458
to something else, I had to build to radio,
165
00:10:54,458 --> 00:10:56,375
I had to build the artificial intelligence.
166
00:10:56,375 --> 00:11:00,330
Not surprisingly, I had to build the vision system,
167
00:11:00,330 --> 00:11:02,541
the navigation system, the manipulators,
168
00:11:02,541 --> 00:11:04,124
the touch screen interfaces, the voice recognition. In fact,
169
00:11:04,124 --> 00:11:10,457
the only thing I could depend on and borrow from other industries
170
00:11:10,457 --> 00:11:12,958
were batteries and motors.
171
00:11:12,958 --> 00:11:17,292
Today, that is no longer the case.
172
00:11:17,292 --> 00:11:21,485
The video game industry and the mobile industry
173
00:11:21,485 --> 00:11:26,122
have solved many, many challenging robot problems.
174
00:11:26,122 --> 00:11:29,542
I can use Siri or Dragon Naturally Dictate
175
00:11:29,542 --> 00:11:33,863
for my speech recognition. I can use wireless communication.
176
00:11:33,863 --> 00:11:39,670
I can use gesture recognition from my kinetic X-Box.
177
00:11:39,670 --> 00:11:42,606
All of these technologies I can assume exist,
178
00:11:42,606 --> 00:11:48,287
and focus the robot industry on navigation, perception
179
00:11:48,287 --> 00:11:50,747
and manipulation.
180
00:11:50,747 --> 00:11:53,517
So, I'd like to give you a little example of
181
00:11:53,517 --> 00:11:56,286
what the state of the art looks like.
182
00:11:56,286 --> 00:12:00,156
In order to make that medical robot work,
183
00:12:00,156 --> 00:12:04,729
the robot has to understand the hospital,
184
00:12:04,729 --> 00:12:08,203
because the doctor isn't going to drive the robot to the patient
185
00:12:08,203 --> 00:12:09,399
he has other things to do.
186
00:12:09,399 --> 00:12:10,287
And in fact,
187
00:12:10,287 --> 00:12:14,504
he may never have been to the hospital before.
188
00:12:14,504 --> 00:12:15,673
He would like to say,
189
00:12:15,673 --> 00:12:18,871
"I want to see patient in room 602"
190
00:12:18,871 --> 00:12:22,704
and the robot should get there all by itself.
191
00:12:22,704 --> 00:12:28,536
so that means that robots have to able to map automatically.
192
00:12:28,536 --> 00:12:31,422
So we've created a robot that does just that.
193
00:12:31,422 --> 00:12:33,870
We drive the robot around the building.
194
00:12:33,870 --> 00:12:36,926
The robot creates a map.
195
00:12:36,926 --> 00:12:42,036
It is registered against a floor plan of the building.
196
00:12:42,036 --> 00:12:47,906
And the robot knows where the room 602 is.
197
00:12:47,906 --> 00:12:50,007
We want the robot to understand and
198
00:12:50,007 --> 00:12:53,704
recognize objects in the room.
199
00:12:53,704 --> 00:12:56,703
So we needed to create a technology
200
00:12:56,703 --> 00:12:58,615
that would allow the robot to say,
201
00:12:58,615 --> 00:13:01,451
"I know what that is".
202
00:13:01,451 --> 00:13:05,924
So we created a system called ViPR,
203
00:13:05,924 --> 00:13:09,598
which recognized exact objects
204
00:13:09,598 --> 00:13:12,829
and copy the exact copies of those objects
205
00:13:12,829 --> 00:13:14,014
regardless of the lighting,
206
00:13:14,014 --> 00:13:17,667
regardless of the orientation of that object,
207
00:13:17,667 --> 00:13:19,671
and do that reliably.
208
00:13:19,671 --> 00:13:21,931
And we do that, and we are able do that
209
00:13:21,931 --> 00:13:25,431
in a very very tiny microprocessor.
210
00:13:25,431 --> 00:13:28,515
So here we have a cell-phone.
211
00:13:28,515 --> 00:13:40,266
So this system will allow a robot to recognize
212
00:13:40,266 --> 00:13:43,326
dollar bills, different objects, as long as they are
213
00:13:43,326 --> 00:13:46,515
exactly like other versions of it.
214
00:13:46,515 --> 00:13:48,466
So that's cool.
215
00:13:48,466 --> 00:13:53,849
But what about generic classes of objects?
216
00:13:53,849 --> 00:13:55,931
When you walked into this auditorium,
217
00:13:55,931 --> 00:13:57,766
you may have never seen a chair that looks
218
00:13:57,766 --> 00:14:00,042
exactly like the one you're sitting in.
219
00:14:00,042 --> 00:14:02,847
And yet, you recognized that's a chair,
220
00:14:02,847 --> 00:14:04,597
and you used it to sit down. Well,
221
00:14:04,597 --> 00:14:09,153
a robot should be able to have that same understanding
222
00:14:09,153 --> 00:14:11,054
of categories of objects
223
00:14:11,054 --> 00:14:13,182
so that it can clean under them.
224
00:14:13,182 --> 00:14:15,091
It can recognize what type of the room it's in,
225
00:14:15,091 --> 00:14:18,896
and have a better awareness of where it is.
226
00:14:18,896 --> 00:14:21,264
So this is a demonstration of a technology
227
00:14:21,264 --> 00:14:23,432
called iSpot that does exactly that.
228
00:14:23,432 --> 00:14:28,439
So what you'll see here is a robot driving through
229
00:14:28,439 --> 00:14:30,508
a building which has a number of different types of
230
00:14:30,508 --> 00:14:33,079
kitchens and different appliances,
231
00:14:33,079 --> 00:14:36,201
and dynamically in real-time, recognizing
232
00:14:36,201 --> 00:14:40,449
the category of the appliance in
233
00:14:40,449 --> 00:14:44,857
that kitchen area automatically.
234
00:14:44,857 --> 00:14:46,623
So if you are a robot and you are doing that mapping
235
00:14:46,623 --> 00:14:51,987
you could also be looking at all the different appliances
236
00:14:51,987 --> 00:14:55,737
and fixtures in that room and be able to tell,
237
00:14:55,737 --> 00:14:57,320
just like your eye could tell,
238
00:14:57,320 --> 00:15:02,373
whether that room is a kitchen or the den or the living room.
239
00:15:02,373 --> 00:15:06,153
so as we get better and better
240
00:15:06,153 --> 00:15:10,447
at making our robots more capable
241
00:15:10,447 --> 00:15:13,321
of understanding their environment and
242
00:15:13,321 --> 00:15:18,737
ultimately able to pick up and lift
243
00:15:18,737 --> 00:15:19,957
and move objects,
244
00:15:19,957 --> 00:15:24,528
we create more and more exciting industries.
245
00:15:24,528 --> 00:15:27,320
And that's very good.
246
00:15:27,320 --> 00:15:30,935
Because 1962,
247
00:15:30,935 --> 00:15:34,820
we saw Rosie the robot and we said
248
00:15:34,820 --> 00:15:39,042
"when are we going to have robots in our lives?"
249
00:15:39,042 --> 00:15:41,478
And 50 years have passed.
250
00:15:41,478 --> 00:15:43,737
And now we're starting
251
00:15:43,737 --> 00:15:46,183
to see some application and some robots that actually
252
00:15:46,183 --> 00:15:49,052
we can afford and bring into our lives.
253
00:15:49,052 --> 00:15:51,988
But if you can compare the number
254
00:15:51,988 --> 00:15:56,090
of robots that are out there today,
255
00:15:56,090 --> 00:15:58,987
and match that up against the need,
256
00:15:58,987 --> 00:16:02,465
match that against the opportunity,
257
00:16:02,465 --> 00:16:05,654
that robots will play in the future,
258
00:16:05,654 --> 00:16:07,820
I believe you will agree with me
259
00:16:07,820 --> 00:16:09,040
when I say:
260
00:16:09,040 --> 00:16:12,375
"we are just about none of the way
261
00:16:12,375 --> 00:16:16,713
in accomplishing what we will accomplish over the next decades."
262
00:16:16,713 --> 00:16:18,849
Thank you very much.
263
00:16:18,849 --> 00:16:20,050
264
00:16:20,050 --> 00:16:21,570
Subtitled by Samantha Hu