1
00:00:02,002 --> 00:00:04,303
It's one of the most
ambitious challenges
2
00:00:04,305 --> 00:00:07,106
in the history of technology.
3
00:00:07,108 --> 00:00:08,741
GILL PRATT:
Ladies and gentlemen,
4
00:00:08,743 --> 00:00:11,677
start your robots!
5
00:00:11,679 --> 00:00:18,050
NARRATOR: The race to build robots
that can save lives in a disaster.
6
00:00:18,052 --> 00:00:21,520
That can even rescue us.
7
00:00:21,522 --> 00:00:25,157
But to do that,
they'll need to climb ladders,
8
00:00:25,159 --> 00:00:28,494
walk over rubble, turn valves.
9
00:00:30,130 --> 00:00:32,064
(cheering)
10
00:00:32,066 --> 00:00:37,903
And no one knows
if it's even possible.
11
00:00:37,905 --> 00:00:41,907
RUSS TEDRAKE: There's so many
things that could go wrong.
12
00:00:41,909 --> 00:00:43,575
It's scary.
13
00:00:43,577 --> 00:00:47,446
VIJAY KUMAR: When you try to
instill human-like intelligence,
14
00:00:47,448 --> 00:00:50,182
human-like manipulation skills
into machines...
15
00:00:51,851 --> 00:00:53,285
it's just very, very hard to do.
16
00:00:53,287 --> 00:00:56,488
NARRATOR: What will it
take for robots to have...
17
00:00:59,192 --> 00:01:01,427
the right stuff?
18
00:01:01,429 --> 00:01:05,197
And are we ready
for the consequences?
19
00:01:05,199 --> 00:01:06,965
SHERRY TURKLE: It's too easy
to look at them and say,
20
00:01:06,967 --> 00:01:07,966
"They're not there yet."
21
00:01:07,968 --> 00:01:10,869
They will get to something
very powerful,
22
00:01:10,871 --> 00:01:14,173
and then you have to say, "Well,
where will we have gotten to?"
23
00:01:14,175 --> 00:01:17,142
AYANNA HOWARD: They might
actually have the capacity
24
00:01:17,144 --> 00:01:18,944
to be smarter than us.
25
00:01:18,946 --> 00:01:24,583
NARRATOR: What impact will
robots have on our future,
26
00:01:24,585 --> 00:01:28,320
on how we work, how we interact,
27
00:01:28,322 --> 00:01:32,724
even how we see ourselves?
28
00:01:32,726 --> 00:01:39,531
"Rise of the Robots,"
right now on NOVA.
29
00:01:56,683 --> 00:02:00,185
Major funding for NOVA
is provided by the following...
30
00:02:00,187 --> 00:02:01,520
with their creations.
31
00:02:01,522 --> 00:02:03,442
ROBOT:
Do you have any questions?
32
00:02:04,691 --> 00:02:09,761
NARRATOR: Especially in Japan, where
bots are starting to cook for us,
33
00:02:09,763 --> 00:02:12,197
do the laundry,
34
00:02:12,199 --> 00:02:15,667
even sing and dance.
35
00:02:15,669 --> 00:02:21,073
Some look so human,
it's hard to tell us apart.
36
00:02:21,075 --> 00:02:25,744
But looks can be deceiving.
37
00:02:25,746 --> 00:02:27,546
People see these humanoid robots
developed in Japan,
38
00:02:27,548 --> 00:02:28,780
all these fancy things,
39
00:02:28,782 --> 00:02:32,718
and they had the expectation
that, "Oh, we have these robots"
40
00:02:32,720 --> 00:02:34,786
and they're going to, you know,
save the world."
41
00:02:34,788 --> 00:02:37,089
But that didn't happen.
42
00:02:37,091 --> 00:02:41,226
NARRATOR:
March 2011.
43
00:02:41,228 --> 00:02:45,364
A devastating earthquake
and tsunami rocked japan.
44
00:02:47,167 --> 00:02:48,467
Explosions and fires
45
00:02:48,469 --> 00:02:50,802
at the Fukushima Daiichi
nuclear power plant
46
00:02:50,804 --> 00:02:54,806
left the area
dangerously radioactive.
47
00:02:54,808 --> 00:03:00,445
Going inside, even for a few
minutes, was life threatening.
48
00:03:02,782 --> 00:03:06,518
It was the perfect time
for Japan's cutting-edge robots
49
00:03:06,520 --> 00:03:09,421
to come to the rescue.
50
00:03:09,423 --> 00:03:11,590
But none did.
51
00:03:11,592 --> 00:03:13,859
PAUL OH:
The real tragedy was
52
00:03:13,861 --> 00:03:16,495
if simply some valves
could have been turned
53
00:03:16,497 --> 00:03:19,631
or some switches
could have been flicked
54
00:03:19,633 --> 00:03:21,567
or hoses attached,
55
00:03:21,569 --> 00:03:26,271
a lot of the meltdown
could have been prevented.
56
00:03:26,273 --> 00:03:30,842
The question was, when robots
were needed the most,
57
00:03:30,844 --> 00:03:34,112
just how come
they weren't effective?
58
00:03:34,114 --> 00:03:38,550
And so I think that led
to a lot of self-reflection.
59
00:03:38,552 --> 00:03:44,189
NARRATOR: After decades of research,
where were our robot heroes?
60
00:03:47,460 --> 00:03:49,695
For generations, science fiction
61
00:03:49,697 --> 00:03:53,232
has portrayed robots
as our loyal servants.
62
00:03:53,234 --> 00:03:56,335
ROBBIE:
Welcome to Altair IV, gentlemen.
63
00:03:56,337 --> 00:03:59,338
I am to transport you
to the residence.
64
00:04:00,940 --> 00:04:03,475
NARRATOR:
But it turns out our imagination
65
00:04:03,477 --> 00:04:07,813
has taken us much further
than our technology.
66
00:04:07,815 --> 00:04:12,117
It's hard to create robots
that can function in our world.
67
00:04:12,119 --> 00:04:14,319
TONY STENTZ:
One of the challenges
68
00:04:14,321 --> 00:04:17,222
with deploying robots
in the real world
69
00:04:17,224 --> 00:04:21,293
is that the real world
is like the wild, wild West.
70
00:04:21,295 --> 00:04:25,163
It's very rich and complex
and very unforgiving.
71
00:04:27,000 --> 00:04:28,467
NARRATOR:
What will it take
72
00:04:28,469 --> 00:04:32,137
for robots to make their way
out of the lab
73
00:04:32,139 --> 00:04:34,573
and into the real world?
74
00:04:36,075 --> 00:04:37,276
We're about to find out.
75
00:04:37,278 --> 00:04:39,645
GILL PRATT:
Ladies and gentlemen,
76
00:04:39,647 --> 00:04:41,246
start your robots!
77
00:04:41,248 --> 00:04:46,718
NARRATOR: At the most ambitious
robotics competition in history.
78
00:04:46,720 --> 00:04:49,688
ANNOUNCER: We welcome you to
the DARPA robotics challenge.
79
00:04:49,690 --> 00:04:53,025
This is where imagination
meets innovation.
80
00:04:53,027 --> 00:04:56,561
You have completed the course!
81
00:04:56,563 --> 00:04:57,929
Whoo!
82
00:04:57,931 --> 00:05:02,934
NARRATOR: In 2013, at a racetrack
on the outskirts of Miami,
83
00:05:02,936 --> 00:05:06,738
DARPA, the research arm of
the U.S. Department of Defense,
84
00:05:06,740 --> 00:05:10,676
challenged 16 teams
from around the world
85
00:05:10,678 --> 00:05:14,212
to build rescue robots that can
help save lives in a disaster.
86
00:05:18,184 --> 00:05:19,918
Very exciting, very exhausted,
87
00:05:19,920 --> 00:05:21,987
very nervous and frustrated
at the same time.
88
00:05:21,989 --> 00:05:24,923
Dealing with robots
is always like that.
89
00:05:24,925 --> 00:05:28,260
NARRATOR: It's the kind of
challenge DARPA is known for:
90
00:05:28,262 --> 00:05:31,229
cutting-edge and high-risk.
91
00:05:35,168 --> 00:05:38,770
DARPA was formed
back in the 1950s
92
00:05:38,772 --> 00:05:40,672
in response
to the launch of Sputnik,
93
00:05:40,674 --> 00:05:43,942
the world's first
artificial satellite
94
00:05:43,944 --> 00:05:45,344
built by the Soviet Union.
95
00:05:46,479 --> 00:05:50,182
Since then,
the agency has spent billions
96
00:05:50,184 --> 00:05:52,684
developing military technology:
97
00:05:52,686 --> 00:05:56,488
advanced weaponry
like stealth technology,
98
00:05:56,490 --> 00:06:00,225
drones, and night vision.
99
00:06:00,227 --> 00:06:03,228
But the spin-offs
from this research
100
00:06:03,230 --> 00:06:05,564
have gone
far beyond the military.
101
00:06:05,566 --> 00:06:09,835
The Internet, GPS,
102
00:06:09,837 --> 00:06:13,138
bionic arms, even Siri,
103
00:06:13,140 --> 00:06:15,107
were all fueled
with DARPA funding.
104
00:06:15,109 --> 00:06:16,608
Cancel golf today.
105
00:06:16,610 --> 00:06:18,143
SIRI:
It's off your calendar.
106
00:06:18,145 --> 00:06:19,145
Good.
107
00:06:20,813 --> 00:06:24,049
NARRATOR: If these researchers
succeed in creating rescue robots,
108
00:06:24,051 --> 00:06:29,421
the same technology
could be used to develop...
109
00:06:29,423 --> 00:06:32,257
robots that take care
of the elderly,
110
00:06:32,259 --> 00:06:34,826
babysit our kids,
111
00:06:34,828 --> 00:06:37,662
clean up after us.
112
00:06:37,664 --> 00:06:44,302
Robots in almost
every facet of life.
113
00:06:44,304 --> 00:06:48,407
But first, can robots
really help out in a disaster?
114
00:06:50,209 --> 00:06:55,847
To find out, DARPA has set up
an obstacle course for bots.
115
00:06:55,849 --> 00:06:59,951
With human operators
controlling their every move...
116
00:06:59,953 --> 00:07:01,319
Yes, we got it!
117
00:07:01,321 --> 00:07:04,523
NARRATOR: the robots
must perform basic tasks,
118
00:07:04,525 --> 00:07:06,291
like opening a door...
119
00:07:06,293 --> 00:07:08,093
All right, go, go, go!
120
00:07:08,095 --> 00:07:09,728
NARRATOR:
turning a valve,
121
00:07:09,730 --> 00:07:12,731
drilling a hole in a wall,
122
00:07:12,733 --> 00:07:15,400
walking over rubble,
123
00:07:15,402 --> 00:07:19,604
even driving a car.
124
00:07:20,840 --> 00:07:25,310
Bot after bot takes their
first steps into the real world.
125
00:07:29,615 --> 00:07:32,617
And there's nothing easy
about it.
126
00:07:34,320 --> 00:07:36,121
NICOLAUS RADFORD:
It's a monumental challenge.
127
00:07:36,123 --> 00:07:37,489
I mean, DARPA calls it
"DARPA hard."
128
00:07:37,491 --> 00:07:39,691
I mean, I almost said it was
"DARPA impossible."
129
00:07:40,993 --> 00:07:43,361
NARRATOR:
Robots fall off ladders.
130
00:07:43,363 --> 00:07:46,565
Some barely move an inch.
131
00:07:46,567 --> 00:07:49,301
They even struggle
to open doors.
132
00:07:51,304 --> 00:07:54,539
The things that a human can do
instinctively, easily,
133
00:07:54,541 --> 00:07:56,575
you don't realize how hard
something like walking is
134
00:07:56,577 --> 00:07:59,144
until you try to reproduce it
in a machine.
135
00:08:01,280 --> 00:08:04,015
PAUL OH: When my baby
daughter took her first step,
136
00:08:04,017 --> 00:08:05,083
did she walk?
137
00:08:05,085 --> 00:08:07,452
No, she took a step
and she fell down, right?
138
00:08:10,756 --> 00:08:13,158
Over time,
she started to walk, right?
139
00:08:13,160 --> 00:08:14,726
And it's kind of like
we're seeing here.
140
00:08:14,728 --> 00:08:15,968
There are a lot of baby steps.
141
00:08:17,997 --> 00:08:20,077
NARRATOR:
After two days of competition...
142
00:08:23,135 --> 00:08:25,604
the results are not impressive.
143
00:08:27,807 --> 00:08:32,010
But this is just the start
of DARPA's grand experiment.
144
00:08:32,012 --> 00:08:34,479
The agency is giving
the roboticists
145
00:08:34,481 --> 00:08:37,215
another chance to get it right,
146
00:08:37,217 --> 00:08:39,951
and they're putting their money
where their mouth is.
147
00:08:41,287 --> 00:08:44,189
The top scoring teams
in this first challenge
148
00:08:44,191 --> 00:08:47,626
are receiving $1 million each
to continue their work.
149
00:08:47,628 --> 00:08:52,564
In 18 months, they will return
for the final challenge
150
00:08:52,566 --> 00:08:56,535
and the chance to win a grand
prize of $2 million.
151
00:08:56,537 --> 00:08:58,003
Yes!
152
00:08:58,005 --> 00:08:59,838
GILL PRATT: It's not clear to me how
well the teams are going to do.
153
00:08:59,840 --> 00:09:03,475
It can't be too hard,
because then everybody fails.
154
00:09:03,477 --> 00:09:04,809
It can't be too easy,
155
00:09:04,811 --> 00:09:06,678
because then it's not
worth doing at all.
156
00:09:06,680 --> 00:09:08,480
But if you make it just right...
157
00:09:08,482 --> 00:09:10,482
Like Goldilocks, right?
158
00:09:11,851 --> 00:09:14,486
I have tried to hit
the sweet spot of difficulty,
159
00:09:14,488 --> 00:09:17,856
but I think risking failure
is the DARPA way.
160
00:09:17,858 --> 00:09:23,028
NARRATOR: Will the challenge push the
technology a giant step forward?
161
00:09:23,030 --> 00:09:28,934
Will robots ever make their way
through our world the way we do?
162
00:09:31,070 --> 00:09:32,537
No!
163
00:09:34,373 --> 00:09:36,107
RODNEY BROOKS:
Back in the early '50s,
164
00:09:36,109 --> 00:09:38,777
Alan Turing, one of the founders
of artificial intelligence,
165
00:09:38,779 --> 00:09:41,746
said that the best thing
we could do was build a robot
166
00:09:41,748 --> 00:09:45,183
with TV cameras for its eyes
and motors to drive its legs
167
00:09:45,185 --> 00:09:47,052
and have it romp
around the countryside
168
00:09:47,054 --> 00:09:49,087
and learn from the real world.
169
00:09:49,089 --> 00:09:51,590
But he decided that was
technologically too hard
170
00:09:51,592 --> 00:09:53,325
back in the '50s,
which it certainly was.
171
00:09:53,327 --> 00:09:57,562
So he said, "Let's leave that
physical interaction until later"
172
00:09:57,564 --> 00:09:59,998
"and let's work
on more abstract problems,
173
00:10:00,000 --> 00:10:02,734
the intelligence
abstract problems."
174
00:10:02,736 --> 00:10:07,238
NARRATOR: The field of artificial
intelligence, or A.I.,
175
00:10:07,240 --> 00:10:10,909
has already built machines
that beat us at chess,
176
00:10:10,911 --> 00:10:13,979
trade stocks
with lightning speed,
177
00:10:13,981 --> 00:10:18,450
and search for anything we want
in an instant.
178
00:10:18,452 --> 00:10:21,886
Yet when it comes to robotics,
179
00:10:21,888 --> 00:10:25,123
progress has been
painfully slow.
180
00:10:25,125 --> 00:10:29,394
Some of the biggest
problems robots face
181
00:10:29,396 --> 00:10:32,430
are things we humans
usually take for granted,
182
00:10:32,432 --> 00:10:37,669
like mobility, manual dexterity,
183
00:10:37,671 --> 00:10:43,208
and the ability to see
and understand our environment.
184
00:10:43,210 --> 00:10:47,178
These are the challenges
the robotics teams will tackle
185
00:10:47,180 --> 00:10:52,017
in the DARPA competition,
beginning with mobility.
186
00:10:53,653 --> 00:10:56,287
What's the best way
to make a machine
187
00:10:56,289 --> 00:10:58,256
that can move through our world?
188
00:10:59,959 --> 00:11:03,395
Does it need to walk on two feet
like us?
189
00:11:03,397 --> 00:11:08,333
Some roboticists think
the answer is yes.
190
00:11:08,335 --> 00:11:12,270
HONG: The shape of the robot is
dictated by what it needs to do.
191
00:11:12,272 --> 00:11:14,539
There's a reason why the step
size in your home is this big,
192
00:11:14,541 --> 00:11:16,474
there's a reason why your
door handle's this high,
193
00:11:16,476 --> 00:11:18,610
because it's designed
for humans to move around.
194
00:11:18,612 --> 00:11:22,213
So unless the robot is
the shape and size of a human,
195
00:11:22,215 --> 00:11:24,482
it won't be able to navigate
and move around
196
00:11:24,484 --> 00:11:26,524
in the environment
designed for humans.
197
00:11:30,990 --> 00:11:36,061
NARRATOR: But getting around on
two feet isn't easy, even for us.
198
00:11:36,063 --> 00:11:39,197
It takes your average infant
almost a year
199
00:11:39,199 --> 00:11:44,636
to go from crawling
to toddling to walking.
200
00:11:44,638 --> 00:11:47,472
What does it take
to give a robot
201
00:11:47,474 --> 00:11:50,408
this intrinsic human ability?
202
00:11:50,410 --> 00:11:53,945
It's one of the biggest problems
facing roboticists today.
203
00:11:57,616 --> 00:11:59,184
In Pensacola, Florida,
204
00:11:59,186 --> 00:12:04,889
at the Institute for Human
Machine Cognition, or IHMC,
205
00:12:04,891 --> 00:12:07,959
one of the teams competing
in the DARPA Robotics Challenge
206
00:12:07,961 --> 00:12:11,996
is hard at work
developing its own software
207
00:12:11,998 --> 00:12:14,933
to run this massive bot
named Atlas,
208
00:12:14,935 --> 00:12:19,104
a 385-pound powerhouse.
209
00:12:19,106 --> 00:12:21,773
DARPA funded the design
and development
210
00:12:21,775 --> 00:12:24,375
of this rescue robot.
211
00:12:24,377 --> 00:12:28,079
Several teams competing
in the challenge
212
00:12:28,081 --> 00:12:30,115
are using this hardware,
213
00:12:30,117 --> 00:12:33,952
but writing unique software
to guide their robot.
214
00:12:36,222 --> 00:12:40,125
You might wonder why Atlas
is so top-heavy.
215
00:12:40,127 --> 00:12:44,095
Most of its oversized head is
packed with cameras and sensors.
216
00:12:44,097 --> 00:12:46,765
And while its feet
may look small,
217
00:12:46,767 --> 00:12:51,402
they're designed to fit
in the kind of places we walk.
218
00:12:51,404 --> 00:12:53,438
JERRY PRATT:
With bipedal walking robots,
219
00:12:53,440 --> 00:12:55,907
there's still a lot of
strategies we have to determine.
220
00:12:57,743 --> 00:13:01,146
And there's no known
textbook solution yet.
221
00:13:01,148 --> 00:13:03,815
It's more art than science
still.
222
00:13:06,018 --> 00:13:08,920
NARRATOR:
For team leader Jerry Pratt,
223
00:13:08,922 --> 00:13:12,791
finding the best way
for a bipedal robot to walk
224
00:13:12,793 --> 00:13:14,959
has not been easy or quick.
225
00:13:14,961 --> 00:13:16,761
PRATT:
Well, I've been working
226
00:13:16,763 --> 00:13:21,199
on bipedal walking robots
since 1994, so 21 years now.
227
00:13:21,201 --> 00:13:25,069
NARRATOR: He started back in
graduate school at the MIT Leg Lab,
228
00:13:25,071 --> 00:13:28,239
where some of the most
bizarre-looking bots
229
00:13:28,241 --> 00:13:33,077
hopped, jumped, and flipped.
230
00:13:33,079 --> 00:13:37,015
But the robots Jerry built
were different
231
00:13:37,017 --> 00:13:39,818
because they walked on two feet.
232
00:13:41,887 --> 00:13:44,889
Using nature as his guide,
233
00:13:44,891 --> 00:13:49,027
Jerry gave his bipeds
hips, knees, and ankles
234
00:13:49,029 --> 00:13:52,564
that mimicked how animals move.
235
00:13:52,566 --> 00:13:56,701
This one,
called Spring Flamingo,
236
00:13:56,703 --> 00:14:00,572
had specially designed motors
that worked a lot like muscles,
237
00:14:00,574 --> 00:14:03,808
varying the amount of power
each joint used.
238
00:14:03,810 --> 00:14:07,612
Its legs and feet worked a bit
like shock absorbers.
239
00:14:07,614 --> 00:14:11,115
They had a little give.
240
00:14:11,117 --> 00:14:13,451
PRATT: That robot was a
really good workhorse.
241
00:14:13,453 --> 00:14:15,753
We had it working
for about three years,
242
00:14:15,755 --> 00:14:17,856
probably walked
about 20 miles or so.
243
00:14:17,858 --> 00:14:21,192
NARRATOR: When it came
to designing a humanoid,
244
00:14:21,194 --> 00:14:23,428
he focused on developing
245
00:14:23,430 --> 00:14:27,198
the perfect combination
of hardware and software
246
00:14:27,200 --> 00:14:30,268
that would enable his biped
to stay upright
247
00:14:30,270 --> 00:14:31,936
as it calculates the best way
248
00:14:31,938 --> 00:14:34,906
to shift its center of mass,
249
00:14:34,908 --> 00:14:38,676
lift its leg, swing it forward,
250
00:14:38,678 --> 00:14:41,379
and put its foot back down
in the right place
251
00:14:41,381 --> 00:14:45,149
and with just the right amount
of pressure.
252
00:14:45,151 --> 00:14:47,952
To make these elaborate
calculations,
253
00:14:47,954 --> 00:14:50,088
the robot needs instructions,
254
00:14:50,090 --> 00:14:53,925
lines of code that tell it
how to move.
255
00:14:53,927 --> 00:14:55,760
These lines of code are written
256
00:14:55,762 --> 00:14:59,230
in a programming language
that resembles English,
257
00:14:59,232 --> 00:15:03,101
but for Atlas to use them,
its onboard computer
258
00:15:03,103 --> 00:15:05,336
has a program
that translates them
259
00:15:05,338 --> 00:15:11,876
into a language a machine
can understand: zeros and ones.
260
00:15:11,878 --> 00:15:16,748
It takes more than two million
lines of code to run Atlas,
261
00:15:16,750 --> 00:15:23,021
500,000 just to put one foot
in front of the other.
262
00:15:23,023 --> 00:15:26,257
It's the interplay
of hardware and software
263
00:15:26,259 --> 00:15:29,794
that keeps this bot on its feet,
264
00:15:29,796 --> 00:15:32,931
an ability that took
hundreds of thousands of years
265
00:15:32,933 --> 00:15:36,301
of human evolution to perfect.
266
00:15:36,303 --> 00:15:37,702
PRATT:
We'll often look
267
00:15:37,704 --> 00:15:40,104
at what strategies does
a human use in order to walk
268
00:15:40,106 --> 00:15:43,241
because the physics in the world
that a human operates in
269
00:15:43,243 --> 00:15:44,609
is the same as a robot.
270
00:15:46,812 --> 00:15:50,114
NARRATOR: But we wouldn't be
able to take a single step
271
00:15:50,116 --> 00:15:53,084
without having
some pretty amazing senses.
272
00:15:55,487 --> 00:15:58,823
When you walk, your eyes detect
the position of your body
273
00:15:58,825 --> 00:16:01,426
relative to the world around it.
274
00:16:01,428 --> 00:16:03,594
At the very same time,
275
00:16:03,596 --> 00:16:06,898
a series of fluid-filled canals
in the inner ear
276
00:16:06,900 --> 00:16:10,468
tells your brain the position
and motion of your head
277
00:16:10,470 --> 00:16:14,272
so you know which way is up.
278
00:16:14,274 --> 00:16:18,109
And a kind of sixth sense
called proprioception
279
00:16:18,111 --> 00:16:20,411
uses your muscles and nerves
280
00:16:20,413 --> 00:16:22,847
to detect where your arms
and legs are
281
00:16:22,849 --> 00:16:25,650
in relation to each other.
282
00:16:25,652 --> 00:16:28,052
All this information
comes together
283
00:16:28,054 --> 00:16:31,622
in a part of the brain
called the cerebellum.
284
00:16:34,460 --> 00:16:38,796
What kind of senses does
Atlas use to walk on two feet?
285
00:16:38,798 --> 00:16:42,934
For eyes,
the bot has a stereo camera
286
00:16:42,936 --> 00:16:45,269
and a couple of fisheye lenses
287
00:16:45,271 --> 00:16:47,472
sticking out the side
of its head,
288
00:16:47,474 --> 00:16:51,576
along with a spinning laser
called LIDAR
289
00:16:51,578 --> 00:16:54,412
that scans everything
in the world around it
290
00:16:54,414 --> 00:16:57,882
and creates a 3-D model
of its environment.
291
00:16:57,884 --> 00:17:01,919
Atlas doesn't have an inner ear
to tell it which way is up.
292
00:17:01,921 --> 00:17:07,258
Instead, it uses
a small cylinder in its butt
293
00:17:07,260 --> 00:17:10,595
that contains gyros
and accelerometers
294
00:17:10,597 --> 00:17:14,565
that tell it where it is
and how it's moving.
295
00:17:14,567 --> 00:17:18,603
In addition, sensors
on each joint tell Atlas
296
00:17:18,605 --> 00:17:22,106
where its limbs are
in relation to each other.
297
00:17:22,108 --> 00:17:25,877
The result is an extraordinary
sense of balance,
298
00:17:25,879 --> 00:17:30,982
allowing Atlas to stand
on one foot like a ballerina...
299
00:17:30,984 --> 00:17:33,584
At least in the lab.
300
00:17:33,586 --> 00:17:36,354
But in the real world,
301
00:17:36,356 --> 00:17:40,892
Atlas's gyros and sensors
aren't always enough.
302
00:17:40,894 --> 00:17:44,762
PRATT: We only have a limited
number of sensors on the robot,
303
00:17:44,764 --> 00:17:47,031
whereas with a human
or an animal,
304
00:17:47,033 --> 00:17:49,667
you have thousands
of little force sensors
305
00:17:49,669 --> 00:17:52,070
on every square inch
of your body.
306
00:17:52,072 --> 00:17:56,874
NARRATOR: Over 100,000 just
on the soles of your feet.
307
00:17:56,876 --> 00:18:00,111
PRATT: You can step on something
and detect what it is,
308
00:18:00,113 --> 00:18:02,080
whereas the robots,
they don't know that.
309
00:18:02,082 --> 00:18:05,316
They just know
they stepped on something.
310
00:18:05,318 --> 00:18:07,351
NARRATOR:
At the final challenge,
311
00:18:07,353 --> 00:18:10,555
if Atlas should step
on something the wrong way,
312
00:18:10,557 --> 00:18:14,725
if the hardware and software
don't work perfectly,
313
00:18:14,727 --> 00:18:18,062
it will fall.
314
00:18:18,064 --> 00:18:21,299
PRATT: We cannot get up
from a fall currently,
315
00:18:21,301 --> 00:18:23,501
and we probably won't
survive a fall.
316
00:18:23,503 --> 00:18:26,237
If we fall during the finals,
317
00:18:26,239 --> 00:18:28,279
that may be the end
of our whole weekend.
318
00:18:32,010 --> 00:18:36,547
NARRATOR: Atlas isn't the only biped
that has trouble staying upright.
319
00:18:36,549 --> 00:18:40,952
Meet HUBO,
designed and programmed
320
00:18:40,954 --> 00:18:43,821
by cousins Paul Oh and Junho Oh.
321
00:18:47,092 --> 00:18:48,826
At the first challenge,
322
00:18:48,828 --> 00:18:53,131
HUBO clearly wasn't ready
for the big, bad world.
323
00:18:53,133 --> 00:18:55,433
PAUL OH:
It's with fondness...
324
00:18:55,435 --> 00:18:58,436
It's also with some pain...
That we think about it.
325
00:18:58,438 --> 00:19:03,040
We went into the challenge
with quite a lot of enthusiasm,
326
00:19:03,042 --> 00:19:04,509
but when we got there,
327
00:19:04,511 --> 00:19:06,591
it was just like
one thing after another.
328
00:19:08,614 --> 00:19:12,049
There were a lot of these
real-world instances
329
00:19:12,051 --> 00:19:15,153
that we did not experience
in the lab.
330
00:19:15,155 --> 00:19:17,421
That was a real awakening.
331
00:19:19,324 --> 00:19:21,092
NARRATOR:
Across the ocean
332
00:19:21,094 --> 00:19:25,463
at the Korea Advanced Institute
of Science and Technology,
333
00:19:25,465 --> 00:19:30,168
Junho Oh is also doing
some soul searching.
334
00:19:52,958 --> 00:19:54,759
NARRATOR:
After the shock wore off,
335
00:19:54,761 --> 00:19:57,728
the cousins agreed
to make a radical change
336
00:19:57,730 --> 00:19:59,931
in the design of their biped.
337
00:19:59,933 --> 00:20:01,432
PAUL HO:
My cousin has come up
338
00:20:01,434 --> 00:20:06,337
with this idea of adding wheels
to HUBO's knees,
339
00:20:06,339 --> 00:20:09,173
as well as casters on its toes.
340
00:20:09,175 --> 00:20:12,043
And I call it
the kneeling prayer mode.
341
00:20:12,045 --> 00:20:14,779
Others like to call it
wheeled mode
342
00:20:14,781 --> 00:20:16,013
or transformer bot mode.
343
00:20:31,430 --> 00:20:33,798
PAUL HO: Humans don't
have wheels on the knees,
344
00:20:33,800 --> 00:20:37,034
but there's no reason why
we can't add that to a robot.
345
00:20:39,338 --> 00:20:41,639
NARRATOR: When the cousins
return for the final challenge,
346
00:20:41,641 --> 00:20:46,344
they'll debut their humanoid,
transformed
347
00:20:46,346 --> 00:20:52,016
with the help
of some clever engineering.
348
00:20:52,018 --> 00:20:55,886
But for a robot to help out
in a disaster,
349
00:20:55,888 --> 00:20:59,190
getting there is just the first
of many challenges.
350
00:21:01,660 --> 00:21:05,162
To assist rescue workers
in the real world,
351
00:21:05,164 --> 00:21:06,664
it needs hands
352
00:21:06,666 --> 00:21:09,000
with the kind of strength
and dexterity it takes
353
00:21:09,002 --> 00:21:11,202
to lift heavy hoses,
354
00:21:11,204 --> 00:21:14,071
drill into walls,
355
00:21:14,073 --> 00:21:17,541
solder and saw.
356
00:21:17,543 --> 00:21:21,846
It needs a pair of these.
357
00:21:21,848 --> 00:21:23,547
OLIVER BROCH:
Our experience of the human hand
358
00:21:23,549 --> 00:21:25,750
is that we can do everything
with it.
359
00:21:25,752 --> 00:21:28,252
We can dig ourselves
through rubble,
360
00:21:28,254 --> 00:21:32,456
we can make really nice
paintings, we can knit.
361
00:21:32,458 --> 00:21:36,761
The human hand really
distinguishes our species
362
00:21:36,763 --> 00:21:38,296
from all other species.
363
00:21:40,565 --> 00:21:43,100
If you look at the way
humans manipulate objects,
364
00:21:43,102 --> 00:21:46,671
we have this instinct
for tactile sensing,
365
00:21:46,673 --> 00:21:48,706
for feeling the world.
366
00:21:48,708 --> 00:21:50,975
Understandably so,
367
00:21:50,977 --> 00:21:53,644
we've had over 200,000 years
of human evolution
368
00:21:53,646 --> 00:21:55,413
to get to this particular point.
369
00:21:56,982 --> 00:22:01,452
NARRATOR: Is it possible to translate
a masterpiece of evolution
370
00:22:01,454 --> 00:22:06,090
into motors, cables, sensors,
and thousands of lines of code?
371
00:22:08,927 --> 00:22:14,765
On the outskirts of London,
behind an unassuming storefront,
372
00:22:14,767 --> 00:22:17,034
a small group
of robot enthusiasts
373
00:22:17,036 --> 00:22:19,704
are building a robotic hand
374
00:22:19,706 --> 00:22:23,207
that they hope could one day
rival our own.
375
00:22:23,209 --> 00:22:24,742
ARMANDO DE LA ROSA T.: When we
first started building a hand,
376
00:22:24,744 --> 00:22:27,478
we actually bought anatomy books
377
00:22:27,480 --> 00:22:29,847
and tried to understand
how the hand works.
378
00:22:29,849 --> 00:22:31,515
We thought it would be good
if we could copy,
379
00:22:31,517 --> 00:22:34,485
as best we can, the human hand.
380
00:22:34,487 --> 00:22:37,188
RICH WALKER: The Shadow Hand was
built with the idea of trying
381
00:22:37,190 --> 00:22:39,423
to get as close as possible
to the human hand,
382
00:22:39,425 --> 00:22:40,691
but as engineers.
383
00:22:41,860 --> 00:22:43,494
NARRATOR:
Like the human hand,
384
00:22:43,496 --> 00:22:47,465
this robotic version has
four fingers and a thumb.
385
00:22:47,467 --> 00:22:50,201
It's about the same size
as the original
386
00:22:50,203 --> 00:22:52,737
and can even shake like one,
387
00:22:52,739 --> 00:22:56,474
but that's where
the similarities end.
388
00:22:56,476 --> 00:23:00,311
We pack a huge amount of sensing
and actuation.
389
00:23:00,313 --> 00:23:03,481
We have 25 joint
position sensors,
390
00:23:03,483 --> 00:23:06,784
nine analog digital converters.
391
00:23:06,786 --> 00:23:08,586
We have two tendons
coming from each joint
392
00:23:08,588 --> 00:23:09,620
to a motor in the forearm.
393
00:23:09,622 --> 00:23:12,156
20 motors in the forearm.
394
00:23:12,158 --> 00:23:14,258
Each motor has
a temperature sensor,
395
00:23:14,260 --> 00:23:17,194
a current sensor,
and two full sensors
396
00:23:17,196 --> 00:23:18,863
so we can tell how hard
the motor is working
397
00:23:18,865 --> 00:23:20,598
and how hard
it's driving the tendons.
398
00:23:20,600 --> 00:23:22,700
We put contact sensing
in the fingertips
399
00:23:22,702 --> 00:23:25,302
so we can tell that
we've touched something.
400
00:23:25,304 --> 00:23:26,904
So there is a wealth
of computing power
401
00:23:26,906 --> 00:23:28,372
just to get something that has
402
00:23:28,374 --> 00:23:31,275
the same set of movements
as your hand.
403
00:23:31,277 --> 00:23:33,277
NARRATOR: And with this
level of dexterity,
404
00:23:33,279 --> 00:23:37,448
it can do a lot more
than card tricks.
405
00:23:37,450 --> 00:23:41,218
DE LA ROSA T.: We have this hand
handling pipettes and lab equipment,
406
00:23:41,220 --> 00:23:43,320
removing the human
from the risk...
407
00:23:43,322 --> 00:23:45,156
For example, people that work
408
00:23:45,158 --> 00:23:48,993
with very, very nasty bacteria
and viruses.
409
00:23:48,995 --> 00:23:53,230
NARRATOR: The Shadow Hand was
designed for delicate tasks,
410
00:23:53,232 --> 00:23:58,002
not for dirty jobs like this.
411
00:23:58,004 --> 00:24:00,271
WALKER: When you look at
where robot hands get used,
412
00:24:00,273 --> 00:24:01,605
you find people who want to do
413
00:24:01,607 --> 00:24:04,074
something delicate
and precise...
414
00:24:05,444 --> 00:24:09,046
and people who want something
big and rugged and solid.
415
00:24:09,048 --> 00:24:11,215
NARRATOR: For the robots
in the DARPA challenge,
416
00:24:11,217 --> 00:24:14,585
strength and precision
are a must.
417
00:24:14,587 --> 00:24:17,521
Finding the balance between them
418
00:24:17,523 --> 00:24:20,958
has been a struggle
for team leader Brett Kennedy,
419
00:24:20,960 --> 00:24:25,996
who developed this four-legged
rescue robot named Robosimian,
420
00:24:25,998 --> 00:24:31,535
here at NASA's Jet Propulsion
Laboratory.
421
00:24:31,537 --> 00:24:33,637
My first intuition is that this
needed to be a very robust hand,
422
00:24:33,639 --> 00:24:36,140
and I was thinking,
"Well, robust means big."
423
00:24:37,576 --> 00:24:41,111
"Why don't you go find out how
big Wilt Chamberlain's hand is
424
00:24:41,113 --> 00:24:45,483
and make a hand that big?"
425
00:24:45,485 --> 00:24:47,351
So if anybody's curious,
426
00:24:47,353 --> 00:24:50,387
Wilt Chamberlain's hand
is very, very big,
427
00:24:50,389 --> 00:24:51,889
and not only is it
very, very big
428
00:24:51,891 --> 00:24:53,891
when you make
a robotic version of it,
429
00:24:53,893 --> 00:24:58,829
it actually cannot deal
with normal human-scale tools.
430
00:24:58,831 --> 00:25:02,266
NARRATOR:
So they modify their design,
431
00:25:02,268 --> 00:25:07,271
making a smaller hand
with just three fingers.
432
00:25:07,273 --> 00:25:10,741
KENNEDY: This does everything we
needed it to in the competition.
433
00:25:10,743 --> 00:25:13,644
It grabbed everything,
the human tools.
434
00:25:13,646 --> 00:25:17,581
So we thought we were
in pretty good shape.
435
00:25:17,583 --> 00:25:20,351
NARRATOR: But then, at
the first challenge,
436
00:25:20,353 --> 00:25:27,057
Robosimian had to open a door,
and things got messy.
437
00:25:27,059 --> 00:25:31,328
So here's an individual finger,
and in that individual finger,
438
00:25:31,330 --> 00:25:35,266
there is an artificial tendon,
a synthetic fiber.
439
00:25:35,268 --> 00:25:37,935
When excessive forces happen,
it snaps.
440
00:25:40,906 --> 00:25:43,307
NARRATOR: When Kennedy
realized Robosimian's fingers
441
00:25:43,309 --> 00:25:47,411
couldn't provide the strength
he was looking for,
442
00:25:47,413 --> 00:25:49,880
he came up with a radically
different idea.
443
00:25:49,882 --> 00:25:52,983
KENNEDY:
So this is the "cam hand."
444
00:25:52,985 --> 00:25:55,853
To do most of the work
that we need to,
445
00:25:55,855 --> 00:25:58,188
a simple hook works just fine
446
00:25:58,190 --> 00:26:00,524
so that we can actually
get a grasp
447
00:26:00,526 --> 00:26:03,294
and we can hold on
to most everything we have to.
448
00:26:03,296 --> 00:26:07,565
The fact of the matter is this
is a very simple, dumb system.
449
00:26:07,567 --> 00:26:09,533
It closes
until it encounters something,
450
00:26:09,535 --> 00:26:12,503
and then that will hold it,
and it'll hold it securely.
451
00:26:15,473 --> 00:26:19,777
NARRATOR:
In their quest to grasp victory,
452
00:26:19,779 --> 00:26:21,879
the teams competing
in the DARPA Challenge
453
00:26:21,881 --> 00:26:24,081
will rely on all kinds of hands.
454
00:26:25,383 --> 00:26:29,520
Many are using this gripper,
designed with three fingers
455
00:26:29,522 --> 00:26:32,690
that can wrap around
a variety of objects
456
00:26:32,692 --> 00:26:35,926
and even pick up
something small.
457
00:26:35,928 --> 00:26:38,796
If we look at tasks like,
for instance, opening a door
458
00:26:38,798 --> 00:26:41,599
or using a drill,
using hand tools and things,
459
00:26:41,601 --> 00:26:45,336
pretty much all these tasks you
can perform using three fingers.
460
00:26:47,939 --> 00:26:49,940
NARRATOR: Robotic grippers
have already made their way
461
00:26:49,942 --> 00:26:55,045
onto the factory floor,
attached to massive robotic arms
462
00:26:55,047 --> 00:26:58,549
that are as robust
as they are precise.
463
00:26:58,551 --> 00:27:02,720
They build cars,
lift heavy boxes,
464
00:27:02,722 --> 00:27:05,155
pack beer,
465
00:27:05,157 --> 00:27:07,725
sort through anything
and everything,
466
00:27:07,727 --> 00:27:11,095
from batteries to pancakes,
467
00:27:11,097 --> 00:27:14,632
doing the kind of jobs
many people consider repetitive
468
00:27:14,634 --> 00:27:16,634
and downright boring.
469
00:27:19,037 --> 00:27:20,971
But some experts fear
470
00:27:20,973 --> 00:27:23,474
as robots move beyond the
factory into the real world,
471
00:27:23,476 --> 00:27:26,777
they'll take on a lot more.
472
00:27:28,980 --> 00:27:31,682
I think that we will lose
at least 20% or 30% of our jobs
473
00:27:31,684 --> 00:27:33,617
over the next 20 years,
and probably more.
474
00:27:33,619 --> 00:27:38,489
Fast food workers are mostly
going to be replaced.
475
00:27:38,491 --> 00:27:40,491
The cashiers at Wal-Mart,
476
00:27:40,493 --> 00:27:43,193
their jobs aren't going to last
that much longer.
477
00:27:43,195 --> 00:27:44,495
The first to go
are going to be drivers.
478
00:27:44,497 --> 00:27:47,898
Uber is spending a lot of money
on driverless cars.
479
00:27:47,900 --> 00:27:49,667
Google is spending
a lot of money.
480
00:27:49,669 --> 00:27:51,235
Now Toyota is.
481
00:27:51,237 --> 00:27:54,004
40 years from now, there are not
going to be a lot of jobs.
482
00:27:54,006 --> 00:27:59,376
NARRATOR: But not eves
the future looks quite so grim.
483
00:27:59,378 --> 00:28:03,614
Over the past century,
while robots and automation
484
00:28:03,616 --> 00:28:05,883
took over
many manufacturing jobs,
485
00:28:05,885 --> 00:28:09,053
other kinds of jobs
have increased.
486
00:28:09,055 --> 00:28:11,922
Some think that trend
will continue.
487
00:28:11,924 --> 00:28:14,625
TERESA GHILARDUCCI: Machines in
all of modern economic history
488
00:28:14,627 --> 00:28:18,996
have helped create jobs,
489
00:28:18,998 --> 00:28:22,099
not taken them away.
490
00:28:22,101 --> 00:28:26,603
Machines are complements
to workers, not substitutes.
491
00:28:26,605 --> 00:28:30,507
They come together and enhance
the productivity of each.
492
00:28:30,509 --> 00:28:31,775
They need each other.
493
00:28:33,545 --> 00:28:37,448
NARRATOR: Today, more and
more jobs require humans
494
00:28:37,450 --> 00:28:41,885
to work side by side with robots
on the assembly line,
495
00:28:41,887 --> 00:28:44,955
programming them
496
00:28:44,957 --> 00:28:46,924
and repairing them.
497
00:28:46,926 --> 00:28:50,861
Technology creates new jobs.
498
00:28:53,164 --> 00:28:56,633
But only for those who have
the skills to adapt.
499
00:28:59,771 --> 00:29:04,241
What happens to the workers
left behind?
500
00:29:04,243 --> 00:29:07,811
GHILARDUCCI: The worker who was
displaced has to be compensated
501
00:29:07,813 --> 00:29:10,447
through retraining
or through a pension.
502
00:29:10,449 --> 00:29:13,484
So that's a social problem,
503
00:29:13,486 --> 00:29:15,619
not a problem rooted
in technology.
504
00:29:17,689 --> 00:29:19,957
MARCUS: People should
already be thinking
505
00:29:19,959 --> 00:29:22,059
about what kind of society
we would want
506
00:29:22,061 --> 00:29:23,301
if not everybody can have jobs.
507
00:29:24,829 --> 00:29:27,498
NARRATOR: Many ethicists say
we should also be thinking
508
00:29:27,500 --> 00:29:31,135
about which jobs we want
to hand over to machines.
509
00:29:31,137 --> 00:29:33,203
RONALD ARKIN: Much of the
things that we are creating
510
00:29:33,205 --> 00:29:36,940
can be used for a whole broad
range of potential applications,
511
00:29:36,942 --> 00:29:40,677
ranging from eldercare
and childcare robots
512
00:29:40,679 --> 00:29:45,816
to healthcare robotic platforms,
even surrogate sex objects.
513
00:29:45,818 --> 00:29:48,452
What is acceptable?
514
00:29:48,454 --> 00:29:51,088
NARRATOR: What impact these
high-tech machines will have
515
00:29:51,090 --> 00:29:56,827
in the workplace
and our homes remains unclear.
516
00:29:56,829 --> 00:29:59,696
But one thing's for sure:
517
00:29:59,698 --> 00:30:02,766
robots are starting to take
their first baby steps
518
00:30:02,768 --> 00:30:06,970
out of the lab
and into the real world,
519
00:30:06,972 --> 00:30:11,942
learning how to manipulate
objects we use every day.
520
00:30:11,944 --> 00:30:16,113
But for a rescue robot
to be truly useful,
521
00:30:16,115 --> 00:30:19,082
there's another hurdle
researchers face,
522
00:30:19,084 --> 00:30:22,319
the toughest one of all:
523
00:30:22,321 --> 00:30:25,489
the challenge
of giving a machine
524
00:30:25,491 --> 00:30:28,859
the ability
to understand its environment...
525
00:30:28,861 --> 00:30:33,263
To give it a brain.
526
00:30:33,265 --> 00:30:36,667
MATT JOHNSON: Our robot doesn't
really have what you'd call a brain.
527
00:30:36,669 --> 00:30:39,369
He's not seeing the world,
he's not perceiving the world.
528
00:30:39,371 --> 00:30:43,273
It's not thinking,
it's not reasoning.
529
00:30:43,275 --> 00:30:44,608
You know, it's pretty dumb.
530
00:30:46,244 --> 00:30:50,347
NARRATOR: In fact, today's
rescue robots are so dumb,
531
00:30:50,349 --> 00:30:56,119
DARPA permits human operators
to guide them, step by step.
532
00:30:56,121 --> 00:30:59,489
The bot waits for instructions
that tell it what to do...
533
00:30:59,491 --> 00:31:04,862
Those zeros and ones
that computers understand.
534
00:31:04,864 --> 00:31:08,565
The instructions travel
through an elaborate pipeline
535
00:31:08,567 --> 00:31:11,001
that connects every piece
of hardware,
536
00:31:11,003 --> 00:31:13,370
every motor and sensor,
537
00:31:13,372 --> 00:31:18,075
to a computer controlled
by a team of human operators.
538
00:31:18,077 --> 00:31:21,011
The operators communicate
with their robots
539
00:31:21,013 --> 00:31:23,013
via a wireless connection,
540
00:31:23,015 --> 00:31:25,949
just like they would
at a real disaster.
541
00:31:25,951 --> 00:31:29,419
And at the final challenge.
542
00:31:29,421 --> 00:31:33,490
These operators
at Carnegie Mellon University
543
00:31:33,492 --> 00:31:37,294
are about to guide their
rescue robot, named CHIMP,
544
00:31:37,296 --> 00:31:40,464
through one of the skills
it needs to master
545
00:31:40,466 --> 00:31:44,868
for the final DARPA Challenge:
turning a valve.
546
00:31:44,870 --> 00:31:47,504
Using six different cameras
547
00:31:47,506 --> 00:31:49,473
located on the front and back
of its head
548
00:31:49,475 --> 00:31:53,310
and a spinning sensor
called LIDAR,
549
00:31:53,312 --> 00:31:59,349
the bot sends data about its
environment to the operators.
550
00:31:59,351 --> 00:32:03,620
A 3-D image of CHIMP's world
appears on their monitors.
551
00:32:03,622 --> 00:32:06,924
CHIMP has no idea
what it's looking at,
552
00:32:06,926 --> 00:32:09,760
so with the click of a mouse,
553
00:32:09,762 --> 00:32:13,397
the humans tell it
exactly where the valve is.
554
00:32:13,399 --> 00:32:16,133
CLARK HAYNES:
Things like recognizing objects,
555
00:32:16,135 --> 00:32:18,302
that's a very hard problem
in robotics.
556
00:32:18,304 --> 00:32:21,204
So we let the human
tell the difference
557
00:32:21,206 --> 00:32:24,341
between a cat and a dog,
a valve and a door.
558
00:32:24,343 --> 00:32:28,645
NARRATOR: Next, they show CHIMP
where to grab the valve,
559
00:32:28,647 --> 00:32:32,382
which way,
and how far to turn it.
560
00:32:32,384 --> 00:32:36,153
But it would be impossibly slow
and impractical
561
00:32:36,155 --> 00:32:39,690
for the operators to tell CHIMP
how to move every joint,
562
00:32:39,692 --> 00:32:44,461
every sensor, every motor
of its complex arm.
563
00:32:44,463 --> 00:32:49,900
This job must be done
by the robot all on its own.
564
00:32:49,902 --> 00:32:53,036
Using a process called
"motion planning,"
565
00:32:53,038 --> 00:32:57,674
CHIMP determines the best path
for its arm to travel.
566
00:32:57,676 --> 00:32:59,676
It shows the operator
what its intentions are,
567
00:32:59,678 --> 00:33:02,245
what its path will be,
how it's going to get there.
568
00:33:02,247 --> 00:33:04,247
There's a plan.
569
00:33:04,249 --> 00:33:05,816
In the end, it's going to be
570
00:33:05,818 --> 00:33:07,651
you say,
"Okay, I approve your plan.
571
00:33:07,653 --> 00:33:10,153
"CHIMP, go ahead and do
your thing," and off it goes.
572
00:33:12,457 --> 00:33:15,792
HAYNES: As we move towards
developing this technology further,
573
00:33:15,794 --> 00:33:18,895
we really want to push
on the robot autonomy
574
00:33:18,897 --> 00:33:22,566
and let CHIMP do more things
on its own,
575
00:33:22,568 --> 00:33:26,203
but we always want to keep this
as a tool for a human.
576
00:33:27,739 --> 00:33:32,542
NARRATOR: In other words, keep the
human in the loop and in control.
577
00:33:34,879 --> 00:33:37,614
But that's not
every team's goal.
578
00:33:39,717 --> 00:33:42,919
At the Massachusetts
Institute of Technology,
579
00:33:42,921 --> 00:33:47,357
Russ Tedrake is taking
a different approach.
580
00:33:47,359 --> 00:33:50,260
He is determined to give
his Atlas robot
581
00:33:50,262 --> 00:33:56,233
as much autonomy as he can
to make it a whole lot smarter.
582
00:33:56,235 --> 00:33:58,769
TEDRAKE:
Our goal as researchers,
583
00:33:58,771 --> 00:34:01,138
especially in this artificial
intelligence lab,
584
00:34:01,140 --> 00:34:03,807
we want to solve
the long-term research questions
585
00:34:03,809 --> 00:34:05,308
about how to make
autonomous robots.
586
00:34:05,310 --> 00:34:06,743
This is the closest
we've ever come
587
00:34:06,745 --> 00:34:09,646
to building an artificial
intelligence machine,
588
00:34:09,648 --> 00:34:11,481
where this humanoid robot
is moving through the world,
589
00:34:11,483 --> 00:34:13,650
solving real problems.
590
00:34:13,652 --> 00:34:15,252
NARRATOR:
At a disaster,
591
00:34:15,254 --> 00:34:18,655
where you can't always count
on a wireless connection,
592
00:34:18,657 --> 00:34:22,659
the smarter the bot gets,
the more it can do on its own.
593
00:34:22,661 --> 00:34:26,530
To create a more
autonomous robot,
594
00:34:26,532 --> 00:34:29,499
Tedrake is developing software
that helps it find
595
00:34:29,501 --> 00:34:34,404
and identify objects
with a bit more independence.
596
00:34:34,406 --> 00:34:38,141
So if the robot is just looking
at my kitchen at home,
597
00:34:38,143 --> 00:34:40,010
there are dishes everywhere,
598
00:34:40,012 --> 00:34:42,212
and you ask the robot
to find a spoon.
599
00:34:42,214 --> 00:34:43,880
That's a really hard question.
600
00:34:43,882 --> 00:34:45,816
If the human just says,
601
00:34:45,818 --> 00:34:47,918
"There's a spoon roughly
over here, click,"
602
00:34:47,920 --> 00:34:50,187
and he just has to look
in a little patch of space
603
00:34:50,189 --> 00:34:53,023
for something that's roughly
the same shape as a spoon,
604
00:34:53,025 --> 00:34:55,992
that takes an extremely hard
problem of object recognition
605
00:34:55,994 --> 00:34:57,661
in a complicated environment
606
00:34:57,663 --> 00:34:59,329
and turns it into a very
simple problem of,
607
00:34:59,331 --> 00:35:01,298
"Okay, I want to look
for spoon-shaped things
608
00:35:01,300 --> 00:35:02,866
in this small region of space."
609
00:35:05,203 --> 00:35:07,204
NARRATOR: While CHIMP
needs its human operators
610
00:35:07,206 --> 00:35:11,675
to tell it exactly where
a valve is and where to grab it,
611
00:35:11,677 --> 00:35:15,779
MIT's Atlas is programmed
to recognize a valve on its own,
612
00:35:15,781 --> 00:35:20,584
figure out the steps it needs
to take as it approaches,
613
00:35:20,586 --> 00:35:25,222
then grab and turn it
with very little human help.
614
00:35:25,224 --> 00:35:27,023
TEDRAKE: We can put it
in an autonomous mode
615
00:35:27,025 --> 00:35:29,292
and basically just watch
the robot execute.
616
00:35:29,294 --> 00:35:30,827
It shows us
what it's about to do.
617
00:35:30,829 --> 00:35:32,095
We could always stop it
618
00:35:32,097 --> 00:35:33,530
if it looked like it was going
to do something wrong,
619
00:35:33,532 --> 00:35:34,698
but when things are going well,
620
00:35:34,700 --> 00:35:39,035
the robot's operating
almost completely autonomously.
621
00:35:39,037 --> 00:35:41,938
There's no real input
622
00:35:41,940 --> 00:35:43,640
coming from us
to the robot at this point.
623
00:35:43,642 --> 00:35:45,742
It's just going on
with its task.
624
00:35:47,712 --> 00:35:50,714
NARRATOR:
A more self-sufficient robot
625
00:35:50,716 --> 00:35:54,651
could potentially help save
more lives in a disaster.
626
00:35:54,653 --> 00:36:00,056
But just how independent
do we want rescue robots to get?
627
00:36:00,058 --> 00:36:03,527
What if an autonomous robot
enters a disaster
628
00:36:03,529 --> 00:36:09,032
and has to decide
who lives and who dies?
629
00:36:09,034 --> 00:36:12,536
It's the kind of moral dilemma
that gave Will Smith nightmares
630
00:36:12,538 --> 00:36:15,739
in the movie I, Robot.
631
00:36:15,741 --> 00:36:17,440
ROBOT:
You are in danger!
632
00:36:17,442 --> 00:36:20,343
NARRATOR: When a bot
chose to save his life
633
00:36:20,345 --> 00:36:23,413
over the life
of an 11-year-old girl.
634
00:36:23,415 --> 00:36:26,082
Save the girl! Save her!
635
00:36:26,084 --> 00:36:28,618
I was the logical choice.
636
00:36:28,620 --> 00:36:32,088
It calculated that I had
a 45% chance of survival.
637
00:36:32,090 --> 00:36:35,125
Sarah only had an 11% chance.
638
00:36:35,127 --> 00:36:39,963
11% is more than enough.
639
00:36:39,965 --> 00:36:41,925
A human being
would have known that.
640
00:36:43,434 --> 00:36:46,169
NARRATOR: While today's
cutting-edge machines,
641
00:36:46,171 --> 00:36:49,940
like the thousands of drones
used by the U.S. military,
642
00:36:49,942 --> 00:36:52,409
still have a human in the loop,
643
00:36:52,411 --> 00:36:55,545
what happens in the future
if they don't?
644
00:36:55,547 --> 00:36:57,747
AYANNA HOWARD: There's
conversations going on right now,
645
00:36:57,749 --> 00:37:00,550
conversations about, what are
the ethical laws of robots?
646
00:37:00,552 --> 00:37:05,722
How far should we really push
this technology?
647
00:37:05,724 --> 00:37:10,193
PETER SINGER: How much autonomy do
we want to give this new technology?
648
00:37:10,195 --> 00:37:12,028
Even the semi-autonomous ones
649
00:37:12,030 --> 00:37:13,997
raise certain
interesting issues,
650
00:37:13,999 --> 00:37:17,667
like, who's to blame
when something goes wrong?
651
00:37:17,669 --> 00:37:18,668
These systems will not be
foolproof,
652
00:37:18,670 --> 00:37:19,669
they will not be perfect.
653
00:37:19,671 --> 00:37:20,971
It's important to remember that.
654
00:37:20,973 --> 00:37:25,842
NARRATOR: How these robotic
technologies, autonomous or not,
655
00:37:25,844 --> 00:37:28,612
will be used on the battlefield
of the future
656
00:37:28,614 --> 00:37:30,914
remains an open question.
657
00:37:32,850 --> 00:37:36,253
But now, just a week
before the final challenge,
658
00:37:36,255 --> 00:37:40,657
the roboticists are starting
to feel the mounting pressure.
659
00:37:40,659 --> 00:37:44,661
Brett Kennedy,
who leads the Robosimian team,
660
00:37:44,663 --> 00:37:46,329
is no exception.
661
00:37:46,331 --> 00:37:47,998
KENNEDY:
I really have no idea
662
00:37:48,000 --> 00:37:49,766
how we're going to place
within the overall field
663
00:37:49,768 --> 00:37:50,967
at the robotics challenge.
664
00:37:50,969 --> 00:37:53,136
All the researchers
that are bringing their teams
665
00:37:53,138 --> 00:37:55,372
are top flight,
666
00:37:55,374 --> 00:37:58,108
so where we end up in that,
I don't know.
667
00:38:00,177 --> 00:38:04,014
NARRATOR: Will Jerry Pratt's years
of research with bipedal robots
668
00:38:04,016 --> 00:38:06,416
finally pay off?
669
00:38:06,418 --> 00:38:10,587
Will Paul Oh and Junho Oh
redeem themselves
670
00:38:10,589 --> 00:38:13,623
after HUBO's poor performance
at the first challenge?
671
00:38:22,967 --> 00:38:26,870
NARRATOR: Is CHIMP's combination
of hardware and software
672
00:38:26,872 --> 00:38:30,307
just what
its human operators need?
673
00:38:30,309 --> 00:38:36,846
Or will more autonomy help
or hinder the team from MIT?
674
00:38:36,848 --> 00:38:39,616
TEDRAKE: There're so many
things that could go wrong.
675
00:38:39,618 --> 00:38:41,618
Geez, the chance that
the robot could break,
676
00:38:41,620 --> 00:38:45,088
or something that worked
99 out of 100 times,
677
00:38:45,090 --> 00:38:46,656
but we get unlucky...
678
00:38:49,927 --> 00:38:51,461
It's scary.
679
00:38:53,130 --> 00:38:55,332
NARRATOR:
June 5, 2015.
680
00:38:55,334 --> 00:38:59,536
Our roboticists meet once again,
681
00:38:59,538 --> 00:39:03,740
this time at the Los Angeles
County Fairgrounds.
682
00:39:05,009 --> 00:39:07,777
Over the next two days,
683
00:39:07,779 --> 00:39:11,414
they'll face off with finalists
from around the world.
684
00:39:11,416 --> 00:39:15,318
Several teams come
from Korea and Japan.
685
00:39:15,320 --> 00:39:18,555
This little robot comes
from Germany.
686
00:39:18,557 --> 00:39:20,957
Most are funded
through government
687
00:39:20,959 --> 00:39:23,126
or corporate sponsorship,
but some bots,
688
00:39:23,128 --> 00:39:26,262
like this odd-looking one
called Cog-Burn,
689
00:39:26,264 --> 00:39:28,832
are funded by the teams
themselves.
690
00:39:33,437 --> 00:39:35,205
Yeah, of course.
691
00:39:35,207 --> 00:39:36,606
Very stiff competition,
692
00:39:36,608 --> 00:39:38,274
looks like there are
a lot of good teams,
693
00:39:38,276 --> 00:39:41,311
probably be a lot that make it
all the way through the course,
694
00:39:41,313 --> 00:39:43,233
so it will come down
to top speed.
695
00:39:46,550 --> 00:39:49,886
NARRATOR: To win, each robot
must perform a series of tasks
696
00:39:49,888 --> 00:39:53,823
a lot like the ones they faced
in the first challenge,
697
00:39:53,825 --> 00:39:57,527
from driving a car,
to drilling a hole in a wall,
698
00:39:57,529 --> 00:40:03,032
to walking over debris,
to tackling a flight of stairs.
699
00:40:03,034 --> 00:40:05,135
The robot that completes
the course
700
00:40:05,137 --> 00:40:07,270
in the shortest time wins.
701
00:40:08,706 --> 00:40:10,440
That puts a lot of pressure
702
00:40:10,442 --> 00:40:13,843
on the operators
to keep their bots moving.
703
00:40:13,845 --> 00:40:16,546
And DARPA has thrown
another wrench into the mix.
704
00:40:18,482 --> 00:40:20,884
Just like at a real disaster,
705
00:40:20,886 --> 00:40:22,886
the power
of the wireless connection
706
00:40:22,888 --> 00:40:26,623
between the operators
and their robots fluctuates.
707
00:40:26,625 --> 00:40:31,361
Sometimes, the robots
will receive a degraded signal,
708
00:40:31,363 --> 00:40:35,365
just bits and pieces of the data
that tells them what to do.
709
00:40:35,367 --> 00:40:40,937
Other times, no data,
no instructions at all.
710
00:40:42,873 --> 00:40:45,909
DARPA is hoping
this will push the roboticists
711
00:40:45,911 --> 00:40:48,611
to develop
more autonomous systems,
712
00:40:48,613 --> 00:40:52,315
bots that can finish a job
without our help.
713
00:40:53,451 --> 00:40:55,318
But what makes
the final challenge
714
00:40:55,320 --> 00:41:00,490
downright nerve-wracking is that
unlike the first competition,
715
00:41:00,492 --> 00:41:02,892
where the bots were tethered,
716
00:41:02,894 --> 00:41:06,529
this time around, there are
no safety lines allowed,
717
00:41:06,531 --> 00:41:09,199
putting these multimillion-
dollar machines
718
00:41:09,201 --> 00:41:11,734
in real jeopardy.
719
00:41:11,736 --> 00:41:13,937
GILL PRATT:
What I was worried about
720
00:41:13,939 --> 00:41:15,472
is that almost none of the teams
721
00:41:15,474 --> 00:41:17,373
had tested their robots
off the safety tether.
722
00:41:17,375 --> 00:41:19,342
We had no idea
when these robots fell
723
00:41:19,344 --> 00:41:22,011
how badly they would break,
724
00:41:22,013 --> 00:41:24,747
so many of the teams
were really scared.
725
00:41:29,787 --> 00:41:33,456
NARRATOR:
Finally, the competition begins.
726
00:41:33,458 --> 00:41:37,427
Each robot has two chances
to run the course.
727
00:41:37,429 --> 00:41:41,064
IHMC starts their first run.
728
00:41:43,567 --> 00:41:46,302
The robot drives
without a hitch.
729
00:41:47,705 --> 00:41:52,141
Walks up to the door with ease.
730
00:41:52,143 --> 00:41:54,410
(audience cheering)
731
00:41:54,412 --> 00:41:56,446
All right, go through the door.
732
00:41:56,448 --> 00:41:58,948
NARRATOR:
All is looking good.
733
00:42:00,918 --> 00:42:03,653
Just two more tasks to go.
734
00:42:03,655 --> 00:42:05,355
Nice.
735
00:42:08,759 --> 00:42:10,393
(audience groans)
736
00:42:12,630 --> 00:42:14,831
No!
737
00:42:17,167 --> 00:42:21,504
NARRATOR:
It's Jerry's worst nightmare.
738
00:42:21,506 --> 00:42:24,674
After years of perfecting
his walking software,
739
00:42:24,676 --> 00:42:28,711
a tiny misstep in the real world
740
00:42:28,713 --> 00:42:32,382
brought one of the most advanced
robots on earth to its knees.
741
00:42:35,152 --> 00:42:36,786
PRATT:
Nobody was really sure
742
00:42:36,788 --> 00:42:38,721
if the Atlas robot
would survive a fall.
743
00:42:38,723 --> 00:42:42,926
You know, we've never had Atlas
fall before today.
744
00:42:42,928 --> 00:42:45,428
NARRATOR: The team worries
the bot can't be repaired
745
00:42:45,430 --> 00:42:47,363
in time for its final run.
746
00:42:52,069 --> 00:42:55,171
For Carnegie Mellon's robot,
CHIMP,
747
00:42:55,173 --> 00:43:00,977
the day is also filled
with ups and downs.
748
00:43:02,212 --> 00:43:03,846
MEYHOFER: When he fell
down through the door,
749
00:43:03,848 --> 00:43:07,750
you know, he's laying there
and we were just, "Oh no."
750
00:43:10,588 --> 00:43:13,690
NARRATOR: Back in the garage,
the operators scramble
751
00:43:13,692 --> 00:43:18,294
to find a way to get the red bot
back up on its treads again.
752
00:43:18,296 --> 00:43:22,031
But the bigger a bot,
the harder it falls.
753
00:43:24,034 --> 00:43:28,171
Smaller robots fall
without breaking
754
00:43:28,173 --> 00:43:32,475
and have truly bizarre ways
of getting back up again.
755
00:43:32,477 --> 00:43:36,512
But they're too small to do
the kind of heavy lifting
756
00:43:36,514 --> 00:43:40,483
rescue robots need to do
at a disaster.
757
00:43:40,485 --> 00:43:43,319
HONG: Once you make it
heavier and larger,
758
00:43:43,321 --> 00:43:45,521
you need stronger motors,
759
00:43:45,523 --> 00:43:47,924
and stronger motor
means heavier motors,
760
00:43:47,926 --> 00:43:52,929
and heavier motors increase
the overall weights.
761
00:43:52,931 --> 00:43:57,800
NARRATOR:
CHIMP weighs over 400 pounds.
762
00:43:57,802 --> 00:44:01,671
That's an awful lot of robot
to lift.
763
00:44:03,674 --> 00:44:07,310
The crowd begins to wonder
if its run is done.
764
00:44:07,312 --> 00:44:09,579
JOHN MARKOFF:
I found it fascinating
765
00:44:09,581 --> 00:44:11,948
to watch the crowd
watching the robots.
766
00:44:11,950 --> 00:44:14,684
This was really
an emotional moment.
767
00:44:14,686 --> 00:44:17,320
They're collections of wires
and gears and motors,
768
00:44:17,322 --> 00:44:23,593
and we were sympathizing
with them at a very basic level.
769
00:44:26,530 --> 00:44:29,932
RRATOR:
Then CHIMP's leg begins to move.
770
00:44:29,934 --> 00:44:33,670
MEYHOFER: This little kid
down in the front screams,
771
00:44:33,672 --> 00:44:35,071
"He's getting up!"
772
00:44:40,644 --> 00:44:42,879
(cheering)
773
00:44:45,616 --> 00:44:49,519
NARRATOR: The bot not only gets
up, it completes all eight tasks,
774
00:44:49,521 --> 00:44:55,091
becoming the first robot
to finish the course.
775
00:45:06,236 --> 00:45:09,105
But not every team is so lucky.
776
00:45:09,107 --> 00:45:13,009
As MIT starts its run,
777
00:45:13,011 --> 00:45:16,279
hoping to show off
its robot's autonomy,
778
00:45:16,281 --> 00:45:19,782
things run amok
in the control room.
779
00:45:19,784 --> 00:45:20,917
We made a simple operator error.
780
00:45:20,919 --> 00:45:22,151
When we went
to get out of the car,
781
00:45:22,153 --> 00:45:25,021
we forgot to turn off
the driving controller,
782
00:45:25,023 --> 00:45:26,956
so the foot tried
to push the throttle
783
00:45:26,958 --> 00:45:29,525
when it was getting out
of the car.
784
00:45:29,527 --> 00:45:31,194
(audience groans)
785
00:45:31,196 --> 00:45:36,466
NARRATOR: In an effort to drive and
get out of the car at the same time,
786
00:45:36,468 --> 00:45:40,837
the robot keels over
and breaks its right arm.
787
00:45:42,673 --> 00:45:44,240
TEDRAKE:
But the operators
788
00:45:44,242 --> 00:45:45,875
were able to recover
789
00:45:45,877 --> 00:45:49,278
and do basically
the entire course left-handed.
790
00:45:49,280 --> 00:45:52,982
NARRATOR: The robot is still
able to do most of the tasks,
791
00:45:52,984 --> 00:45:56,853
but it faces
one insurmountable problem:
792
00:45:56,855 --> 00:46:00,189
the team has programmed it
to use two hands
793
00:46:00,191 --> 00:46:03,126
to pick up the drill
and turn it on.
794
00:46:03,128 --> 00:46:08,564
It can't do that
when one of its hands is broken.
795
00:46:08,566 --> 00:46:11,300
Without completing
all the tasks,
796
00:46:11,302 --> 00:46:15,938
team MIT has no chance
of winning.
797
00:46:21,111 --> 00:46:23,913
A weary team IHMC
798
00:46:23,915 --> 00:46:28,951
has worked through the night
trying to fix its robot.
799
00:46:28,953 --> 00:46:30,086
Somehow, the robot's
still working,
800
00:46:30,088 --> 00:46:32,488
but the fall we took
did something with the robot
801
00:46:32,490 --> 00:46:36,692
where he's kind of
all messed up a little bit.
802
00:46:36,694 --> 00:46:39,829
NARRATOR: There's something
wrong with its LIDAR,
803
00:46:39,831 --> 00:46:43,166
the system that scans
the robot's environment.
804
00:46:43,168 --> 00:46:45,168
PRATT: The operator is going
to have to adjust on the fly
805
00:46:45,170 --> 00:46:49,138
and compensate in his head
for the errors in our sensors.
806
00:46:49,140 --> 00:46:52,708
Are we really leaning that much?
807
00:46:52,710 --> 00:46:54,110
Our strategy is still
what it was yesterday:
808
00:46:54,112 --> 00:46:55,478
eight points or bust.
809
00:47:00,250 --> 00:47:03,953
NARRATOR: The time has
come for their final run.
810
00:47:07,024 --> 00:47:08,257
All right!
811
00:47:08,259 --> 00:47:11,561
NARRATOR: They make it
through all the tasks
812
00:47:11,563 --> 00:47:13,362
and approach the dreaded debris
813
00:47:13,364 --> 00:47:16,732
where they fell
on their first run.
814
00:47:16,734 --> 00:47:19,335
All right, you gotta
swing the right first.
815
00:47:19,337 --> 00:47:22,004
NARRATOR:
But this time, they succeed.
816
00:47:25,375 --> 00:47:30,012
The operator guides their robot
all the way up the stairs,
817
00:47:30,014 --> 00:47:33,049
finishing the course
in record time.
818
00:47:33,051 --> 00:47:35,451
Yes!
819
00:47:38,155 --> 00:47:39,789
You know, having a human
in the loop,
820
00:47:39,791 --> 00:47:41,624
you've got a supercomputer
up here.
821
00:47:41,626 --> 00:47:44,627
Humans can adapt
to just about anything.
822
00:47:44,629 --> 00:47:46,562
Eight points confirmed!
823
00:47:46,564 --> 00:47:49,866
NARRATOR: The team is now in
the running for first place.
824
00:47:52,469 --> 00:47:54,403
But a one-of-a-kind bot
825
00:47:54,405 --> 00:47:56,739
could still give them
a serious run for their money.
826
00:47:56,741 --> 00:48:02,178
Cousins Paul and Junho are each
competing with their own robot,
827
00:48:02,180 --> 00:48:06,482
doubling the chances that
one of their humanoids on wheels
828
00:48:06,484 --> 00:48:11,220
will win the day.
829
00:48:11,222 --> 00:48:13,856
Paul's team makes it
as far as the drilling task
830
00:48:13,858 --> 00:48:20,463
until the drill gets stuck,
overheats, and shuts down.
831
00:48:20,465 --> 00:48:23,833
But cousin Junho's team
moves on.
832
00:48:23,835 --> 00:48:26,669
(applause)
833
00:48:26,671 --> 00:48:29,705
Their robot completes
all the manual tasks
834
00:48:29,707 --> 00:48:34,911
by rolling on its knees
or standing on its feet.
835
00:48:36,813 --> 00:48:37,880
(cheering)
836
00:48:40,751 --> 00:48:44,854
But when it approaches
its final task, the stairs,
837
00:48:44,856 --> 00:48:46,489
it suddenly stops.
838
00:48:47,557 --> 00:48:51,127
Back in the garage,
its human handlers
839
00:48:51,129 --> 00:48:53,496
are double- and triple-checking
the instructions
840
00:48:53,498 --> 00:48:56,666
they're about to send
their robot.
841
00:48:56,668 --> 00:49:01,404
A mistake here will cost them
the competition.
842
00:49:13,417 --> 00:49:15,651
NARRATOR:
Their job is done.
843
00:49:15,653 --> 00:49:19,522
All they can do now
is watch and wait.
844
00:49:24,661 --> 00:49:28,464
Finally, HUBO starts
to climb the stairs
845
00:49:28,466 --> 00:49:31,600
unlike any human on earth,
846
00:49:31,602 --> 00:49:36,205
with its head facing forward
and its feet facing back.
847
00:49:36,207 --> 00:49:40,576
The humanoid on wheels
ufinishes all eight tasks,
848
00:49:40,578 --> 00:49:43,479
faster than any other robot.
849
00:49:43,481 --> 00:49:48,250
Junho Oh's robot HUBO
wins the day.
850
00:49:48,252 --> 00:49:54,090
IHMC comes in second.
851
00:49:54,092 --> 00:49:57,727
Team CHIMP, from Carnegie
Mellon, comes in third.
852
00:50:01,231 --> 00:50:05,901
These bots may have moved
slowly, but just like a toddler,
853
00:50:05,903 --> 00:50:11,707
they're taking baby steps
into our world.
854
00:50:11,709 --> 00:50:13,275
BROOKS:
We tend to think
855
00:50:13,277 --> 00:50:15,311
the world today is what
it's going to be like
856
00:50:15,313 --> 00:50:16,545
in ten years and twenty years.
857
00:50:16,547 --> 00:50:18,314
But if we look back ten years
what we have,
858
00:50:18,316 --> 00:50:19,882
look back twenty years
what we have,
859
00:50:19,884 --> 00:50:21,784
the world has been
totally transformed.
860
00:50:21,786 --> 00:50:23,319
It's really hard for us
861
00:50:23,321 --> 00:50:25,554
to imagine how different
it's going to be.
862
00:50:25,556 --> 00:50:29,658
VIJAY KUMAR: In some ways,
the change in the field
863
00:50:29,660 --> 00:50:31,894
has been very incremental,
but in other ways,
864
00:50:31,896 --> 00:50:33,856
it's been completely
unpredictable.
865
00:50:35,132 --> 00:50:37,500
SHERRY TURKLE: It's too easy
to look at them and say,
866
00:50:37,502 --> 00:50:38,701
"Oh, they're not there yet."
867
00:50:38,703 --> 00:50:41,904
Well, they will get
to something very powerful.
868
00:50:41,906 --> 00:50:46,008
And then you have to say, "Well,
where will we have gotten to?"
869
00:50:47,644 --> 00:50:51,614
NARRATOR: In fact, a few
months after the finals,
870
00:50:51,616 --> 00:50:55,885
Boston Dynamics, the makers
of the Atlas robot,
871
00:50:55,887 --> 00:50:57,386
took this updated version
for a stroll
872
00:50:57,388 --> 00:51:00,289
in a snow-covered woods.
873
00:51:00,291 --> 00:51:02,024
While it stumbles,
874
00:51:02,026 --> 00:51:05,461
it quickly recovers its balance
just as we would.
875
00:51:06,930 --> 00:51:09,999
TURKLE: I think we just
need to approach this
876
00:51:10,001 --> 00:51:14,970
with our eyes open
and understand our vulnerability
877
00:51:14,972 --> 00:51:20,509
to a technology that we are
on the cusp of creating.
878
00:51:20,511 --> 00:51:24,013
NARRATOR: There's no question
that developing rescue robots
879
00:51:24,015 --> 00:51:28,484
with the potential to save lives
makes a lot of sense.
880
00:51:30,353 --> 00:51:35,825
But the potential for other
applications remains unclear.
881
00:51:37,094 --> 00:51:44,934
The time is now to think
about their role in our lives,
882
00:51:44,936 --> 00:51:49,572
as we face
the "Rise of the Robots."
883
00:52:12,496 --> 00:52:15,264
NEIL ARMSTRONG:
That's one small step for man...
884
00:52:26,543 --> 00:52:29,345
This NOVA program
is available on DVD.
885
00:52:29,347 --> 00:52:34,550
To order, visit shopPBS.org,
or call 1-800-PLAY-PBS.
886
00:52:34,552 --> 00:52:37,186
NOVA is also available
for download on iTunes.