1
00:00:35,245 --> 00:00:38,182
In my lab, we build
autonomous aerial robots

2
00:00:38,183 --> 00:00:40,423
like the one you see flying here.

3
00:00:42,575 --> 00:00:45,750
Unlike the commercially available drones
that you can buy today,

4
00:00:46,738 --> 00:00:49,704
this robot doesn't have any GPS on board.

5
00:00:51,160 --> 00:00:52,376
So without GPS,

6
00:00:52,400 --> 00:00:55,400
it's hard for robots like this
to determine their position.

7
00:00:56,953 --> 00:01:01,311
This robot uses onboard sensors,
cameras and laser scanners,

8
00:01:01,389 --> 00:01:03,521
to scan the environment.

9
00:01:03,522 --> 00:01:06,522
It detects features from the environment,

10
00:01:07,199 --> 00:01:09,918
and it determines where it is
relative to those features,

11
00:01:09,951 --> 00:01:12,021
using a method of triangulation.

12
00:01:13,642 --> 00:01:17,094
And then it can assemble
all these features into a map,

13
00:01:17,189 --> 00:01:19,267
like you see behind me.

14
00:01:19,714 --> 00:01:23,593
And this map then allows the robot
to understand where the obstacles are

15
00:01:23,660 --> 00:01:26,070
and navigate in a collision-free manner.

16
00:01:27,570 --> 00:01:29,491
What I want to show you next

17
00:01:30,025 --> 00:01:33,025
is a set of experiments
we did inside our laboratory,

18
00:01:33,050 --> 00:01:36,050
where this robot was able
to go for longer distances.

19
00:01:37,240 --> 00:01:42,378
So here you'll see, on the top right,
what the robot sees with the camera.

20
00:01:42,403 --> 00:01:43,619
And on the main screen...

21
00:01:43,644 --> 00:01:46,100
And of course this is sped up
by a factor of four...

22
00:01:46,125 --> 00:01:48,792
On the main screen you'll see
the map that it's building.

23
00:01:48,817 --> 00:01:52,654
So this is a high-resolution map
of the corridor around our laboratory.

24
00:01:52,921 --> 00:01:55,257
And in a minute
you'll see it enter our lab,

25
00:01:55,282 --> 00:01:58,138
which is recognizable
by the clutter that you see.

26
00:01:58,186 --> 00:01:59,202
(Laughter)

27
00:01:59,227 --> 00:02:01,234
But the main point I want to convey to you

28
00:02:01,259 --> 00:02:04,724
is that these robots are capable
of building high-resolution maps

29
00:02:04,762 --> 00:02:07,258
at five centimeters resolution,

30
00:02:07,361 --> 00:02:11,528
allowing somebody who is outside the lab,
or outside the building

31
00:02:11,553 --> 00:02:14,712
to deploy these
without actually going inside,

32
00:02:14,809 --> 00:02:17,809
and trying to infer
what happens inside the building.

33
00:02:19,637 --> 00:02:21,877
Now there's one problem
with robots like this.

34
00:02:22,901 --> 00:02:25,101
The first problem is it's pretty big.

35
00:02:25,492 --> 00:02:27,172
Because it's big, it's heavy.

36
00:02:27,917 --> 00:02:30,917
And these robots consume
about 100 watts per pound.

37
00:02:31,756 --> 00:02:34,036
And this makes for
a very short mission life.

38
00:02:35,254 --> 00:02:36,710
The second problem

39
00:02:36,735 --> 00:02:40,668
is that these robots have onboard sensors
that end up being very expensive...

40
00:02:40,796 --> 00:02:43,796
A laser scanner, a camera
and the processors.

41
00:02:44,525 --> 00:02:47,525
That drives up the cost of this robot.

42
00:02:48,741 --> 00:02:51,068
So we asked ourselves a question:

43
00:02:51,422 --> 00:02:54,988
what consumer product
can you buy in an electronics store

44
00:02:55,221 --> 00:03:00,901
that is inexpensive, that's lightweight,
that has sensing onboard and computation?

45
00:03:03,350 --> 00:03:06,006
And we invented the flying phone.

46
00:03:06,053 --> 00:03:07,989
(Laughter)

47
00:03:08,014 --> 00:03:13,881
So this robot uses a Samsung Galaxy
smartphone that you can buy off the shelf,

48
00:03:13,906 --> 00:03:17,837
and all you need is an app that you
can download from our app store.

49
00:03:18,206 --> 00:03:22,303
And you can see this robot
reading the letters, "TED" in this case,

50
00:03:22,390 --> 00:03:25,326
looking at the corners
of the "T" and the "E"

51
00:03:25,351 --> 00:03:28,793
and then triangulating off of that,
flying autonomously.

52
00:03:29,989 --> 00:03:33,354
That joystick is just there
to make sure if the robot goes crazy,

53
00:03:33,379 --> 00:03:34,795
Giuseppe can kill it.

54
00:03:34,820 --> 00:03:36,460
(Laughter)

55
00:03:37,951 --> 00:03:40,951
In addition to building
these small robots,

56
00:03:41,600 --> 00:03:46,180
we also experiment with aggressive
behaviors, like you see here.

57
00:03:46,943 --> 00:03:51,240
So this robot is now traveling
at two to three meters per second,

58
00:03:52,080 --> 00:03:55,080
pitching and rolling aggressively
as it changes direction.

59
00:03:55,529 --> 00:03:59,607
The main point is we can have
smaller robots that can go faster

60
00:03:59,632 --> 00:04:02,592
and then travel in these
very unstructured environments.

61
00:04:04,223 --> 00:04:06,279
And in this next video,

62
00:04:06,304 --> 00:04:11,688
just like you see this bird, an eagle,
gracefully coordinating its wings,

63
00:04:12,620 --> 00:04:16,238
its eyes and feet
to grab prey out of the water,

64
00:04:16,940 --> 00:04:18,836
our robot can go fishing, too.

65
00:04:18,860 --> 00:04:20,356
(Laughter)

66
00:04:20,380 --> 00:04:24,434
In this case, this is a Philly cheesesteak
hoagie that it's grabbing out of thin air.

67
00:04:24,459 --> 00:04:26,860
(Laughter)

68
00:04:27,180 --> 00:04:30,510
So you can see this robot
going at about three meters per second,

69
00:04:30,510 --> 00:04:34,987
which is faster than walking speed,
coordinating its arms, its claws

70
00:04:35,660 --> 00:04:39,680
and its flight with split-second timing
to achieve this maneuver.

71
00:04:41,414 --> 00:04:42,630
In another experiment,

72
00:04:42,655 --> 00:04:45,655
I want to show you
how the robot adapts its flight

73
00:04:46,285 --> 00:04:48,661
to control its suspended payload,

74
00:04:48,686 --> 00:04:52,372
whose length is actually larger
than the width of the window.

75
00:04:52,949 --> 00:04:54,645
So in order to accomplish this,

76
00:04:54,670 --> 00:04:57,943
it actually has to pitch
and adjust the altitude

77
00:04:58,390 --> 00:05:00,710
and swing the payload through.

78
00:05:06,134 --> 00:05:08,430
But of course we want
to make these even smaller,

79
00:05:08,455 --> 00:05:11,455
and we're inspired
in particular by honeybees.

80
00:05:11,480 --> 00:05:14,480
So if you look at honeybees,
and this is a slowed down video,

81
00:05:14,798 --> 00:05:18,889
they're so small,
the inertia is so lightweight...

82
00:05:19,460 --> 00:05:20,636
(Laughter)

83
00:05:20,660 --> 00:05:21,810
that they don't care...

84
00:05:21,810 --> 00:05:24,220
They bounce off my hand, for example.

85
00:05:24,220 --> 00:05:27,220
This is a little robot
that mimics the honeybee behavior.

86
00:05:27,862 --> 00:05:29,078
And smaller is better,

87
00:05:29,103 --> 00:05:32,675
because along with the small size
you get lower inertia.

88
00:05:32,701 --> 00:05:34,237
Along with lower inertia...

89
00:05:34,460 --> 00:05:37,316
(Robot buzzing, laughter)

90
00:05:37,340 --> 00:05:40,156
along with lower inertia,
you're resistant to collisions.

91
00:05:40,180 --> 00:05:41,900
And that makes you more robust.

92
00:05:43,300 --> 00:05:45,956
So just like these honeybees,
we build small robots.

93
00:05:45,980 --> 00:05:48,980
And this particular one
is only 25 grams in weight.

94
00:05:49,141 --> 00:05:51,301
It consumes only six watts of power.

95
00:05:51,701 --> 00:05:54,237
And it can travel
up to six meters per second.

96
00:05:54,262 --> 00:05:56,598
So if I normalize that to its size,

97
00:05:56,623 --> 00:05:59,623
it's like a Boeing 787 traveling
ten times the speed of sound.

98
00:06:03,229 --> 00:06:05,325
(Laughter)

99
00:06:05,350 --> 00:06:07,270
And I want to show you an example.

100
00:06:08,340 --> 00:06:13,095
This is probably the first planned mid-air
collision, at one-twentieth normal speed.

101
00:06:13,318 --> 00:06:16,176
These are going at a relative speed
of two meters per second,

102
00:06:16,201 --> 00:06:18,681
and this illustrates the basic principle.

103
00:06:19,445 --> 00:06:24,163
The two-gram carbon fiber cage around it
prevents the propellers from entangling,

104
00:06:24,278 --> 00:06:29,289
but essentially the collision is absorbed
and the robot responds to the collisions.

105
00:06:30,059 --> 00:06:32,619
And so small also means safe.

106
00:06:32,900 --> 00:06:34,916
In my lab, as we developed these robots,

107
00:06:34,940 --> 00:06:36,560
we start off with these big robots

108
00:06:36,584 --> 00:06:39,396
and then now we're down
to these small robots.

109
00:06:39,420 --> 00:06:42,750
And if you plot a histogram
of the number of Band-Aids we've ordered

110
00:06:42,900 --> 00:06:45,476
in the past, that sort of tailed off now.

111
00:06:45,500 --> 00:06:47,460
Because these robots are really safe.

112
00:06:48,860 --> 00:06:51,316
The small size has some disadvantages,

113
00:06:51,340 --> 00:06:55,516
and nature has found a number of ways
to compensate for these disadvantages.

114
00:06:56,060 --> 00:07:00,116
The basic idea is they aggregate
to form large groups, or swarms.

115
00:07:01,020 --> 00:07:04,884
So, similarly, in our lab,
we try to create artificial robot swarms.

116
00:07:05,020 --> 00:07:06,401
And this is quite challenging

117
00:07:06,425 --> 00:07:10,035
because now you have to think
about networks of robots.

118
00:07:10,260 --> 00:07:11,556
And within each robot,

119
00:07:11,580 --> 00:07:16,606
you have to think about the interplay
of sensing, communication, computation...

120
00:07:17,420 --> 00:07:22,022
And this network then becomes
quite difficult to control and manage.

121
00:07:23,260 --> 00:07:26,260
So from nature we take away
three organizing principles

122
00:07:26,580 --> 00:07:29,580
that essentially allow us
to develop our algorithms.

123
00:07:30,740 --> 00:07:35,124
The first idea is that robots
need to be aware of their neighbors.

124
00:07:35,482 --> 00:07:38,723
They need to be able to sense
and communicate with their neighbors.

125
00:07:39,740 --> 00:07:42,396
So this video illustrates the basic idea.

126
00:07:42,420 --> 00:07:43,716
You have four robots...

127
00:07:43,741 --> 00:07:47,428
One of the robots has actually been
hijacked by a human operator, literally.

128
00:07:48,917 --> 00:07:51,156
But because the robots
interact with each other,

129
00:07:51,180 --> 00:07:52,836
they sense their neighbors,

130
00:07:52,860 --> 00:07:54,156
they essentially follow.

131
00:07:54,180 --> 00:07:59,309
And here there's a single person
able to lead this network of followers.

132
00:08:01,700 --> 00:08:06,675
So again, it's not because all the robots
know where they're supposed to go.

133
00:08:06,780 --> 00:08:10,717
It's because they're just reacting
to the positions of their neighbors.

134
00:08:13,420 --> 00:08:16,420
(Laughter)

135
00:08:17,980 --> 00:08:22,317
So the next experiment illustrates
the second organizing principle.

136
00:08:24,620 --> 00:08:27,620
And this principle has to do
with the principle of anonymity.

137
00:08:29,100 --> 00:08:32,100
Here the key idea is that

138
00:08:33,419 --> 00:08:37,200
the robots are agnostic
to the identities of their neighbors.

139
00:08:38,140 --> 00:08:40,756
They're asked to form a circular shape,

140
00:08:40,780 --> 00:08:44,052
and no matter how many robots
you introduce into the formation,

141
00:08:44,100 --> 00:08:46,676
or how many robots you pull out,

142
00:08:46,700 --> 00:08:49,700
each robot is simply
reacting to its neighbor.

143
00:08:49,859 --> 00:08:54,357
It's aware of the fact that it needs
to form the circular shape,

144
00:08:54,860 --> 00:08:56,636
but collaborating with its neighbors

145
00:08:56,660 --> 00:08:59,660
it forms the shape
without central coordination.

146
00:09:02,020 --> 00:09:04,436
Now if you put these ideas together,

147
00:09:04,461 --> 00:09:08,356
the third idea is that we
essentially give these robots

148
00:09:08,380 --> 00:09:12,086
mathematical descriptions
of the shape they need to execute.

149
00:09:12,700 --> 00:09:15,700
And these shapes can be varying
as a function of time,

150
00:09:16,220 --> 00:09:20,524
and you'll see these robots
start from a circular formation,

151
00:09:20,740 --> 00:09:23,995
change into a rectangular formation,
stretch into a straight line,

152
00:09:24,020 --> 00:09:25,395
back into an ellipse.

153
00:09:25,419 --> 00:09:29,032
And they do this with the same
kind of split-second coordination

154
00:09:29,060 --> 00:09:32,060
that you see in natural swarms, in nature.

155
00:09:33,980 --> 00:09:36,116
So why work with swarms?

156
00:09:36,140 --> 00:09:39,736
Let me tell you about two applications
that we are very interested in.

157
00:09:40,195 --> 00:09:42,901
The first one has to do with agriculture,

158
00:09:42,926 --> 00:09:46,052
which is probably the biggest problem
that we're facing worldwide.

159
00:09:46,779 --> 00:09:48,035
As you well know,

160
00:09:48,060 --> 00:09:51,687
one in every seven persons
in this earth is malnourished.

161
00:09:51,725 --> 00:09:55,224
Most of the land that we can cultivate
has already been cultivated.

162
00:09:56,026 --> 00:09:59,605
And the efficiency of most systems
in the world is improving,

163
00:09:59,630 --> 00:10:02,932
but our production system
efficiency is actually declining.

164
00:10:03,114 --> 00:10:06,864
And that's mostly because of water
shortage, crop diseases, climate change

165
00:10:07,473 --> 00:10:08,993
and a couple of other things.

166
00:10:09,466 --> 00:10:10,946
So what can robots do?

167
00:10:11,306 --> 00:10:15,895
Well, we adopt an approach that's
called Precision Farming in the community.

168
00:10:15,954 --> 00:10:20,946
And the basic idea is that we fly
aerial robots through orchards,

169
00:10:21,203 --> 00:10:24,203
and then we build
precision models of individual plants.

170
00:10:24,935 --> 00:10:26,706
So just like personalized medicine,

171
00:10:26,731 --> 00:10:31,021
while you might imagine wanting
to treat every patient individually,

172
00:10:31,450 --> 00:10:34,450
what we'd like to do is build
models of individual plants

173
00:10:35,202 --> 00:10:38,887
and then tell the farmer
what kind of inputs every plant needs...

174
00:10:39,607 --> 00:10:43,863
The inputs in this case being water,
fertilizer and pesticide.

175
00:10:45,516 --> 00:10:49,328
Here you'll see robots
traveling through an apple orchard,

176
00:10:49,353 --> 00:10:51,609
and in a minute you'll see
two of its companions

177
00:10:51,634 --> 00:10:53,444
doing the same thing on the left side.

178
00:10:54,100 --> 00:10:57,315
And what they're doing is essentially
building a map of the orchard.

179
00:10:57,780 --> 00:11:02,275
Within the map is a map
of every plant in this orchard.

180
00:11:02,420 --> 00:11:04,076
(Robot buzzing)

181
00:11:04,100 --> 00:11:05,996
Let's see what those maps look like.

182
00:11:06,020 --> 00:11:10,715
In the next video, you'll see the cameras
that are being used on this robot.

183
00:11:10,740 --> 00:11:13,740
On the top-left is essentially
a standard color camera.

184
00:11:14,838 --> 00:11:17,838
On the left-center is an infrared camera.

185
00:11:18,174 --> 00:11:21,791
And on the bottom-left
is a thermal camera.

186
00:11:22,014 --> 00:11:25,447
And on the main panel, you're seeing
a three-dimensional reconstruction

187
00:11:25,447 --> 00:11:30,348
of every tree in the orchard
as the sensors fly right past the trees.

188
00:11:32,989 --> 00:11:36,593
Armed with information like this,
we can do several things.

189
00:11:37,477 --> 00:11:41,812
The first and possibly the most important
thing we can do is very simple:

190
00:11:41,837 --> 00:11:44,277
count the number of fruits on every tree.

191
00:11:44,726 --> 00:11:49,436
By doing this, you tell the farmer
how many [fruits] she has in every tree

192
00:11:49,580 --> 00:11:52,580
and allow her to estimate
the yield in the orchard,

193
00:11:53,860 --> 00:11:56,700
optimizing the production
chain downstream.

194
00:11:57,940 --> 00:11:59,556
The second thing we can do

195
00:11:59,580 --> 00:12:03,645
is take models of plants, construct
three-dimensional reconstructions,

196
00:12:04,100 --> 00:12:06,636
and from that estimate the canopy size,

197
00:12:06,660 --> 00:12:10,327
and then correlate the canopy size
to the amount of leaf area on every plant.

198
00:12:10,460 --> 00:12:12,636
And this is called the leaf area index.

199
00:12:12,660 --> 00:12:14,596
So if you know this leaf area index,

200
00:12:14,620 --> 00:12:19,810
you essentially have a measure of how much
photosynthesis is possible in every plant,

201
00:12:20,100 --> 00:12:22,980
which again tells you
how healthy each plant is.

202
00:12:25,620 --> 00:12:29,795
By combining visual
and infrared information,

203
00:12:30,355 --> 00:12:33,355
we can also compute indices such as NDVI.

204
00:12:33,780 --> 00:12:36,596
And in this particular case,
you can essentially see

205
00:12:36,620 --> 00:12:39,620
there are some crops that are
not doing as well as other crops.

206
00:12:39,660 --> 00:12:43,363
This is easily discernible from imagery,

207
00:12:44,340 --> 00:12:46,556
not just visual imagery but combining

208
00:12:46,580 --> 00:12:49,356
both visual imagery and infrared imagery.

209
00:12:50,180 --> 00:12:51,516
And then lastly,

210
00:12:51,540 --> 00:12:55,798
one thing we're interested in doing is
detecting the early onset of chlorosis...

211
00:12:55,980 --> 00:12:57,476
And this is an orange tree...

212
00:12:57,500 --> 00:13:00,060
Which is essentially seen
by yellowing of leaves.

213
00:13:00,085 --> 00:13:03,688
But robots flying overhead
can easily spot this autonomously

214
00:13:03,935 --> 00:13:06,871
and then report to the farmer
that he or she has a problem

215
00:13:06,896 --> 00:13:08,416
in this section of the orchard.

216
00:13:10,100 --> 00:13:12,796
Systems like this can really help,

217
00:13:12,820 --> 00:13:18,619
and we're projecting yields
that can improve by about ten percent

218
00:13:18,660 --> 00:13:21,750
and, more importantly, decrease
the amount of inputs such as water

219
00:13:21,755 --> 00:13:24,755
by 25 percent by using
aerial robot swarms.

220
00:13:25,740 --> 00:13:29,021
A second application area
is in first response.

221
00:13:29,038 --> 00:13:31,427
This is a picture of the Penn campus,

222
00:13:31,427 --> 00:13:33,918
actually south of the Penn campus,
the South Bank.

223
00:13:33,942 --> 00:13:36,116
I want you to imagine a setting

224
00:13:36,141 --> 00:13:39,130
where there might be an emergency call
from a building,

225
00:13:39,155 --> 00:13:40,377
a 911 call.

226
00:13:40,528 --> 00:13:44,837
By the time the police get there,
it might take a valuable 5-10 minutes.

227
00:13:44,879 --> 00:13:47,012
But imagine now, robots respond.

228
00:13:47,300 --> 00:13:49,426
And you have a whole swarm of them,

229
00:13:49,442 --> 00:13:52,974
flying to the scene autonomously,
just triggered by a 911 call

230
00:13:52,990 --> 00:13:54,451
or by a dispacher.

231
00:13:55,295 --> 00:13:57,449
By the way, if someone is here
from the FAA,

232
00:13:57,449 --> 00:13:59,557
this was actually shot in South America.

233
00:13:59,588 --> 00:14:01,434
(Laughter)

234
00:14:01,474 --> 00:14:03,541
So, robots arrive at the scene,

235
00:14:04,858 --> 00:14:07,882
and they're all equipped
with downward facing cameras,

236
00:14:08,232 --> 00:14:10,299
and they can monitor the scene.

237
00:14:10,343 --> 00:14:12,057
And they do this autonomously,

238
00:14:12,081 --> 00:14:16,370
so by the time a human first responder
or a police officer gets there,

239
00:14:16,609 --> 00:14:19,466
they have access
to all kinds of information.

240
00:14:19,649 --> 00:14:22,649
So on the top left,
you see the central screen

241
00:14:22,673 --> 00:14:24,473
that a dispacher might see,

242
00:14:24,498 --> 00:14:27,236
which is telling her
where the robots are flying

243
00:14:27,244 --> 00:14:29,632
and how they're encircling the building.

244
00:14:29,657 --> 00:14:33,196
And the robots are autonomously deciding
which ingress poins

245
00:14:33,292 --> 00:14:35,492
should be assigned to what robot.

246
00:14:36,412 --> 00:14:40,205
On the top right, you essentialy see
images from the robots

247
00:14:40,220 --> 00:14:42,220
being assembled into a mosaic.

248
00:14:42,529 --> 00:14:46,085
Which again, gives the first responder
some idea

249
00:14:46,346 --> 00:14:48,743
of what activity is going on at the scene.

250
00:14:48,768 --> 00:14:51,860
And on the bottom, you can see
a three-dimensional reconstruction

251
00:14:51,860 --> 00:14:53,800
that we developed on the fly.

252
00:14:55,103 --> 00:14:57,302
In addition to working outside buidlings,

253
00:14:57,302 --> 00:14:59,847
we're also interested
in going inside buidlings,

254
00:14:59,855 --> 00:15:03,172
and I want to show you
an experiment we did three years ago

255
00:15:03,450 --> 00:15:07,188
where our aerial robot -
one exactly like this one -

256
00:15:07,768 --> 00:15:10,234
is collaborating with a ground robot,

257
00:15:10,268 --> 00:15:13,268
in this case it's actually hitching a ride
with a ground robot,

258
00:15:13,300 --> 00:15:16,300
because it's programmed to be lazy,
to save power.

259
00:15:16,453 --> 00:15:19,453
So, as it goes up,
the two travel in tandem,

260
00:15:19,585 --> 00:15:22,585
and this is a collapsed building
after an earthquake,

261
00:15:22,624 --> 00:15:25,609
this is shortly after the 2011 earthquake
in Fukushima.

262
00:15:26,763 --> 00:15:29,763
When the robots get stuck
in front of a collapsed doorway,

263
00:15:29,784 --> 00:15:33,458
our tobot takes off
and is able to fly over the obstacles

264
00:15:33,950 --> 00:15:35,350
around the obstacles,

265
00:15:35,498 --> 00:15:38,736
and generate a three-dimensional map,
in this case of a bookcase.

266
00:15:39,072 --> 00:15:41,436
And it's able to see
what's on the other side.

267
00:15:41,458 --> 00:15:45,513
Something fairly simple,
but it's hard to do with ground robots,

268
00:15:45,530 --> 00:15:49,045
and certainly hard to do with humans
when there's potential for harm.

269
00:15:50,151 --> 00:15:54,085
So these two robots
are collaboratively building these maps,

270
00:15:54,617 --> 00:15:57,188
and, again,
when the first responders come,

271
00:15:57,213 --> 00:15:59,480
they can be quick with these maps.

272
00:15:59,630 --> 00:16:02,630
So let me show you
what some of these maps look like.

273
00:16:02,720 --> 00:16:04,545
So this is a three storey building,

274
00:16:04,570 --> 00:16:07,537
the seventh, eighth and what remains
of the ninth floor on top,

275
00:16:07,562 --> 00:16:11,164
and we were able to construct this map
using this team of two robots,

276
00:16:11,188 --> 00:16:13,455
operating in tandem, autonomously.

277
00:16:14,032 --> 00:16:17,528
However, this experiment took us
two and a half hours to complete.

278
00:16:18,569 --> 00:16:20,751
Now, no first responder
is going to give you

279
00:16:20,776 --> 00:16:23,124
two and a half hours
to do this experiment,

280
00:16:23,149 --> 00:16:25,228
before he or she wants to rush in.

281
00:16:25,252 --> 00:16:27,283
They might give you
two and a half minutes,

282
00:16:27,292 --> 00:16:29,684
you'll be lucky if you get
two and a half seconds.

283
00:16:29,700 --> 00:16:31,874
But now that's where robot swarms
come in.

284
00:16:31,883 --> 00:16:34,533
Imagine if you could send in
a hundred of these robots,

285
00:16:34,538 --> 00:16:36,711
like the little ones
that we were just flying,

286
00:16:36,711 --> 00:16:39,625
and imagine they went in
and generated maps like this

287
00:16:39,649 --> 00:16:42,649
well before humans actually arrived
on the scene.

288
00:16:43,189 --> 00:16:46,189
And that's the vision
we're working towards.

289
00:16:47,570 --> 00:16:48,837
So, let me conclude

290
00:16:49,994 --> 00:16:52,993
with a movie -
a Warner brothers movie -

291
00:16:53,228 --> 00:16:56,228
of an upcoming -
right next in your theatre,

292
00:16:57,138 --> 00:16:59,082
The Swarm!
The Swarm is coming!

293
00:16:59,098 --> 00:17:01,934
And I love this poster,
actually if you've seen the movie,

294
00:17:01,990 --> 00:17:04,057
you're probably dating yourself

295
00:17:04,738 --> 00:17:07,689
if you have not seen the movie,
I encourage you not to see it,

296
00:17:07,689 --> 00:17:08,836
it's a terrible movie,

297
00:17:08,836 --> 00:17:09,885
(Laughter)

298
00:17:09,885 --> 00:17:13,347
It's about killer bees, attacking men
and killing them and so on.

299
00:17:13,363 --> 00:17:16,490
But everything about this poster is true,
which is why I like it.

300
00:17:16,498 --> 00:17:19,369
"Its size is immeasurable" -
I hope I've convinced you

301
00:17:19,369 --> 00:17:21,237
that "its power is limitless",

302
00:17:21,307 --> 00:17:22,917
and even this last bit is true,

303
00:17:22,934 --> 00:17:25,862
"its enemy is man",
the technology is here today

304
00:17:25,862 --> 00:17:27,905
and it's people like us that are standing

305
00:17:27,905 --> 00:17:30,605
between this technology
and its applications.

306
00:17:30,625 --> 00:17:33,625
The swarm is coming,
this is not science fiction,

307
00:17:34,109 --> 00:17:36,309
in fact, this is what lies ahead.

308
00:17:36,734 --> 00:17:42,299
Lastly, I want you to applaud
the people who actually create the future,

309
00:17:42,647 --> 00:17:47,328
Yash Mulgaonkar, Sikang Liu
and Giuseppe Loianno,

310
00:17:47,416 --> 00:17:50,416
who are responsible for the three
demonstrations that you saw.

311
00:17:51,103 --> 00:17:52,279
Thank you.

312
00:17:52,328 --> 00:17:55,328
(Applause)