1
00:00:01,125 --> 00:00:05,018
In 2015, 25 teams from around the world

2
00:00:05,042 --> 00:00:07,768
competed to build robots
for disaster response

3
00:00:07,792 --> 00:00:09,643
that could perform a number of tasks,

4
00:00:09,667 --> 00:00:11,309
such as using a power tool,

5
00:00:11,333 --> 00:00:13,018
working on uneven terrain

6
00:00:13,042 --> 00:00:14,292
and driving a car.

7
00:00:15,292 --> 00:00:17,333
That all sounds impressive, and it is,

8
00:00:18,458 --> 00:00:21,250
but look at the body
of the winning robot, HUBO.

9
00:00:22,208 --> 00:00:24,518
Here, HUBO is trying to get out of a car,

10
00:00:24,542 --> 00:00:26,268
and keep in mind,

11
00:00:26,292 --> 00:00:28,059
the video is sped up three times.

12
00:00:28,083 --> 00:00:31,708
(Laughter)

13
00:00:32,750 --> 00:00:35,726
HUBO, from team KAIST out of Korea,
is a state-of-the-art robot

14
00:00:35,750 --> 00:00:37,518
with impressive capabilities,

15
00:00:37,542 --> 00:00:39,684
but this body doesn't look
all that different

16
00:00:39,708 --> 00:00:41,661
from robots we've seen a few decades ago.

17
00:00:42,625 --> 00:00:45,042
If you look at the other robots
in the competition,

18
00:00:46,000 --> 00:00:49,101
their movements also still look,
well, very robotic.

19
00:00:49,125 --> 00:00:51,351
Their bodies are complex
mechanical structures

20
00:00:51,375 --> 00:00:52,976
using rigid materials

21
00:00:53,000 --> 00:00:56,601
such as metal and traditional
rigid electric motors.

22
00:00:56,625 --> 00:00:58,226
They certainly weren't designed

23
00:00:58,250 --> 00:01:01,309
to be low-cost, safe near people

24
00:01:01,333 --> 00:01:04,184
and adaptable to unpredictable challenges.

25
00:01:04,208 --> 00:01:06,976
We've made good progress
with the brains of robots,

26
00:01:07,000 --> 00:01:08,792
but their bodies are still primitive.

27
00:01:10,625 --> 00:01:12,643
This is my daughter Nadia.

28
00:01:12,667 --> 00:01:14,393
She's only five years old

29
00:01:14,417 --> 00:01:16,893
and she can get out of the car
way faster than HUBO.

30
00:01:16,917 --> 00:01:18,684
(Laughter)

31
00:01:18,708 --> 00:01:21,101
She can also swing around
on monkey bars with ease,

32
00:01:21,125 --> 00:01:24,268
much better than any current
human-like robot could do.

33
00:01:24,292 --> 00:01:25,643
In contrast to HUBO,

34
00:01:25,667 --> 00:01:29,393
the human body makes extensive use
of soft and deformable materials

35
00:01:29,417 --> 00:01:31,351
such as muscle and skin.

36
00:01:31,375 --> 00:01:33,726
We need a new generation of robot bodies

37
00:01:33,750 --> 00:01:37,851
that is inspired by the elegance,
efficiency and by the soft materials

38
00:01:37,875 --> 00:01:39,417
of the designs found in nature.

39
00:01:40,292 --> 00:01:44,518
And indeed, this has become
the key idea of a new field of research

40
00:01:44,542 --> 00:01:46,476
called soft robotics.

41
00:01:46,500 --> 00:01:49,393
My research group
and collaborators around the world

42
00:01:49,417 --> 00:01:53,059
are using soft components
inspired by muscle and skin

43
00:01:53,083 --> 00:01:55,893
to build robots with agility and dexterity

44
00:01:55,917 --> 00:01:57,559
that comes closer and closer

45
00:01:57,583 --> 00:02:00,875
to the astonishing capabilities
of the organisms found in nature.

46
00:02:02,458 --> 00:02:06,143
I've always been particularly inspired
by biological muscle.

47
00:02:06,167 --> 00:02:08,309
Now, that's not surprising.

48
00:02:08,333 --> 00:02:11,934
I'm also Austrian, and I know that I sound
a bit like Arnie, the Terminator.

49
00:02:11,958 --> 00:02:14,976
(Laughter)

50
00:02:15,000 --> 00:02:17,976
Biological muscle
is a true masterpiece of evolution.

51
00:02:18,000 --> 00:02:19,559
It can heal after damage

52
00:02:19,583 --> 00:02:22,018
and it's tightly integrated
with sensory neurons

53
00:02:22,042 --> 00:02:24,167
for feedback on motion
and the environment.

54
00:02:25,083 --> 00:02:27,976
It can contract fast enough
to power the high-speed wings

55
00:02:28,000 --> 00:02:29,268
of a hummingbird;

56
00:02:29,292 --> 00:02:31,976
it can grow strong enough
to move an elephant;

57
00:02:32,000 --> 00:02:35,559
and it's adaptable enough
to be used in the extremely versatile arms

58
00:02:35,583 --> 00:02:36,893
of an octopus,

59
00:02:36,917 --> 00:02:40,250
an animal that can squeeze
its entire body through tiny holes.

60
00:02:41,250 --> 00:02:45,143
Actuators are for robots
what muscles are for animals:

61
00:02:45,167 --> 00:02:47,018
key components of the body

62
00:02:47,042 --> 00:02:49,583
that enable movement
and interaction with the world.

63
00:02:50,667 --> 00:02:53,143
So if we could build soft actuators,

64
00:02:53,167 --> 00:02:54,601
or artificial muscles,

65
00:02:54,625 --> 00:02:56,226
that are as versatile, adaptable

66
00:02:56,250 --> 00:02:58,809
and could have the same performance
as the real thing,

67
00:02:58,833 --> 00:03:00,768
we could build almost any type of robot

68
00:03:00,792 --> 00:03:02,167
for almost any type of use.

69
00:03:03,167 --> 00:03:06,351
Not surprisingly,
people have tried for many decades

70
00:03:06,375 --> 00:03:09,268
to replicate the astonishing
capabilities of muscle,

71
00:03:09,292 --> 00:03:10,625
but it's been really hard.

72
00:03:12,583 --> 00:03:14,059
About 10 years ago,

73
00:03:14,083 --> 00:03:16,809
when I did my PhD back in Austria,

74
00:03:16,833 --> 00:03:18,934
my colleagues and I rediscovered

75
00:03:18,958 --> 00:03:23,101
what is likely one of the very first
publications on artificial muscle,

76
00:03:23,125 --> 00:03:24,976
published in 1880.

77
00:03:25,000 --> 00:03:28,143
"On the shape and volume changes
of dielectric bodies

78
00:03:28,167 --> 00:03:29,684
caused by electricity,"

79
00:03:29,708 --> 00:03:32,809
published by German physicist
Wilhelm Röntgen.

80
00:03:32,833 --> 00:03:35,875
Most of you know him
as the discoverer of the X-ray.

81
00:03:37,292 --> 00:03:39,851
Following his instructions,
we used a pair of needles.

82
00:03:39,875 --> 00:03:41,829
We connected it to a high-voltage source,

83
00:03:41,833 --> 00:03:44,309
and we placed it near
a transparent piece of rubber

84
00:03:44,313 --> 00:03:46,397
that was prestretched
onto a plastic frame.

85
00:03:47,333 --> 00:03:49,018
When we switched on the voltage,

86
00:03:49,042 --> 00:03:50,351
the rubber deformed,

87
00:03:50,375 --> 00:03:53,559
and just like our biceps flexes our arm,

88
00:03:53,583 --> 00:03:56,226
the rubber flexed the plastic frame.

89
00:03:56,250 --> 00:03:57,559
It looks like magic.

90
00:03:57,583 --> 00:03:59,934
The needles don't even touch the rubber.

91
00:03:59,958 --> 00:04:02,476
Now, having two such needles
is not a practical way

92
00:04:02,500 --> 00:04:04,643
of operating artificial muscles,

93
00:04:04,667 --> 00:04:07,809
but this amazing experiment
got me hooked on the topic.

94
00:04:07,833 --> 00:04:10,851
I wanted to create new ways
to build artificial muscles

95
00:04:10,875 --> 00:04:13,684
that would work well
for real-world applications.

96
00:04:13,708 --> 00:04:17,434
For the next years, I worked
on a number of different technologies

97
00:04:17,458 --> 00:04:18,767
that all showed promise,

98
00:04:18,791 --> 00:04:21,833
but they all had remaining challenges
that are hard to overcome.

99
00:04:22,833 --> 00:04:24,434
In 2015,

100
00:04:24,458 --> 00:04:26,851
when I started my own lab at CU Boulder,

101
00:04:26,875 --> 00:04:29,101
I wanted to try an entirely new idea.

102
00:04:29,125 --> 00:04:32,018
I wanted to combine
the high speed and efficiency

103
00:04:32,042 --> 00:04:34,018
of electrically driven actuators

104
00:04:34,042 --> 00:04:36,792
with the versatility
of soft, fluidic actuators.

105
00:04:37,458 --> 00:04:38,768
Therefore, I thought,

106
00:04:38,792 --> 00:04:42,143
maybe I can try using
really old science in a new way.

107
00:04:42,167 --> 00:04:43,934
The diagram you see here

108
00:04:43,958 --> 00:04:46,518
shows an effect called Maxwell stress.

109
00:04:46,542 --> 00:04:48,018
When you take two metal plates

110
00:04:48,042 --> 00:04:50,226
and place them in a container
filled with oil,

111
00:04:50,250 --> 00:04:51,893
and then switch on a voltage,

112
00:04:51,917 --> 00:04:55,851
the Maxwell stress forces the oil
up in between the two plates,

113
00:04:55,875 --> 00:04:57,351
and that's what you see here.

114
00:04:57,375 --> 00:04:58,726
So the key idea was,

115
00:04:58,750 --> 00:05:01,684
can we use this effect to push around oil

116
00:05:01,708 --> 00:05:04,518
contained in soft stretchy structures?

117
00:05:04,542 --> 00:05:06,893
And indeed, this worked surprisingly well,

118
00:05:06,917 --> 00:05:09,601
quite honestly,
much better than I expected.

119
00:05:09,625 --> 00:05:11,809
Together with my
outstanding team of students,

120
00:05:11,833 --> 00:05:13,726
we used this idea as a starting point

121
00:05:13,750 --> 00:05:17,893
to develop a new technology
called HASEL artificial muscles.

122
00:05:17,917 --> 00:05:21,018
HASELs are gentle enough
to pick up a raspberry

123
00:05:21,042 --> 00:05:22,333
without damaging it.

124
00:05:24,542 --> 00:05:27,750
They can expand and contract
like real muscle.

125
00:05:29,625 --> 00:05:32,083
And they can be operated
faster than the real thing.

126
00:05:32,958 --> 00:05:35,601
They can also be scaled up
to deliver large forces.

127
00:05:35,625 --> 00:05:38,976
Here you see them lifting
a gallon filled with water.

128
00:05:39,000 --> 00:05:40,934
They can be used to drive a robotic arm,

129
00:05:40,958 --> 00:05:43,042
and they can even
self-sense their position.

130
00:05:45,292 --> 00:05:48,000
HASELs can be used
for very precise movement,

131
00:05:49,292 --> 00:05:52,143
but they can also deliver
very fluidic, muscle-like movement

132
00:05:52,167 --> 00:05:55,042
and bursts of power
to shoot up a ball into the air.

133
00:05:57,375 --> 00:05:58,958
When submerged in oil,

134
00:06:00,667 --> 00:06:04,208
HASEL artificial muscles
can be made invisible.

135
00:06:07,875 --> 00:06:09,792
So how do HASEL artificial muscles work?

136
00:06:10,542 --> 00:06:12,226
You might be surprised.

137
00:06:12,250 --> 00:06:15,434
They're based on very inexpensive,
easily available materials.

138
00:06:15,458 --> 00:06:17,726
You can even try, and I recommend it,

139
00:06:17,750 --> 00:06:19,083
the main principle at home.

140
00:06:19,875 --> 00:06:22,643
Take a few Ziploc bags
and fill them with olive oil.

141
00:06:22,667 --> 00:06:24,875
Try to push out air bubbles
as much as you can.

142
00:06:25,583 --> 00:06:28,684
Now take a glass plate
and place it on one side of the bag.

143
00:06:28,708 --> 00:06:30,917
When you press down,
you see the bag contract.

144
00:06:31,792 --> 00:06:34,458
Now the amount of contraction
is easy to control.

145
00:06:35,167 --> 00:06:37,976
When you take a small weight,
you get a small contraction.

146
00:06:38,000 --> 00:06:41,125
With a medium weight,
we get a medium contraction.

147
00:06:42,125 --> 00:06:45,018
And with a large weight,
you get a large contraction.

148
00:06:45,042 --> 00:06:48,476
Now for HASELs, the only change
is to replace the force of your hand

149
00:06:48,500 --> 00:06:51,518
or the weight with an electrical force.

150
00:06:51,542 --> 00:06:56,893
HASEL stands for "hydraulically amplified
self-healing electrostatic actuators."

151
00:06:56,917 --> 00:07:00,101
Here you see a geometry
called Peano-HASEL actuators,

152
00:07:00,125 --> 00:07:02,417
one of many possible designs.

153
00:07:03,083 --> 00:07:06,684
Again, you take a flexible polymer
such as our Ziploc bag,

154
00:07:06,708 --> 00:07:09,643
you fill it with an insulating liquid,
such as olive oil,

155
00:07:09,667 --> 00:07:11,434
and now, instead of the glass plate,

156
00:07:11,458 --> 00:07:14,250
you place an electrical conductor
on one side of the pouch.

157
00:07:14,958 --> 00:07:17,601
To create something
that looks more like a muscle fiber,

158
00:07:17,625 --> 00:07:19,601
you can connect a few pouches together

159
00:07:19,625 --> 00:07:21,768
and attached a weight on one side.

160
00:07:21,792 --> 00:07:23,167
Next, we apply voltage.

161
00:07:24,083 --> 00:07:27,393
Now, the electric field
starts acting on the liquid.

162
00:07:27,417 --> 00:07:29,101
It displaces the liquid,

163
00:07:29,125 --> 00:07:31,125
and it forces the muscle to contract.

164
00:07:32,708 --> 00:07:35,184
Here you see a completed
Peano-HASEL actuator

165
00:07:35,208 --> 00:07:39,101
and how it expands and contracts
when voltage is applied.

166
00:07:39,125 --> 00:07:40,434
Viewed from the side,

167
00:07:40,458 --> 00:07:43,684
you can really see those pouches
take a more cylindrical shape,

168
00:07:43,708 --> 00:07:45,500
such as we saw with the Ziploc bags.

169
00:07:46,500 --> 00:07:49,726
We can also place a few
such muscle fibers next to each other

170
00:07:49,750 --> 00:07:52,309
to create something that looks
even more like a muscle

171
00:07:52,333 --> 00:07:54,768
that also contracts and expands
in cross section.

172
00:07:54,792 --> 00:07:58,143
These HASELs here are lifting a weight
that's about 200 times heavier

173
00:07:58,167 --> 00:07:59,458
than their own weight.

174
00:08:00,667 --> 00:08:04,143
Here you see one of our newest designs,
called quadrant donut HASELs

175
00:08:04,167 --> 00:08:05,768
and how they expand and contract.

176
00:08:05,792 --> 00:08:08,875
They can be operated incredibly fast,
reaching superhuman speeds.

177
00:08:11,333 --> 00:08:14,309
They are even powerful enough
to jump off the ground.

178
00:08:14,333 --> 00:08:15,667
(Laughter)

179
00:08:16,917 --> 00:08:20,434
Overall, HASELs show promise
to become the first technology

180
00:08:20,458 --> 00:08:23,976
that matches or exceeds the performance
of biological muscle

181
00:08:24,000 --> 00:08:27,143
while being compatible
with large-scale manufacturing.

182
00:08:27,167 --> 00:08:30,309
This is also a very young technology.
We are just getting started.

183
00:08:30,333 --> 00:08:33,101
We have many ideas how to
drastically improve performance,

184
00:08:33,125 --> 00:08:36,976
using new materials and new designs
to reach a level of performance

185
00:08:37,000 --> 00:08:41,332
beyond biological muscle and also beyond
traditional rigid electric motors.

186
00:08:42,250 --> 00:08:45,643
Moving towards more complex designs
of HASEL for bio-inspired robotics,

187
00:08:45,667 --> 00:08:47,434
here you see our artificial scorpion

188
00:08:47,458 --> 00:08:49,184
that can use its tail to hunt prey,

189
00:08:49,208 --> 00:08:50,726
in this case, a rubber balloon.

190
00:08:50,750 --> 00:08:52,393
(Laughter)

191
00:08:52,417 --> 00:08:54,518
Going back to our initial inspiration,

192
00:08:54,542 --> 00:08:57,351
the versatility of octopus arms
and elephant trunks,

193
00:08:57,375 --> 00:09:00,143
we are now able to build
soft continuum actuators

194
00:09:00,167 --> 00:09:03,292
that come closer and closer
to the capabilities of the real thing.

195
00:09:05,833 --> 00:09:08,726
I am most excited
about the practical applications

196
00:09:08,750 --> 00:09:10,559
of HASEL artificial muscles.

197
00:09:10,583 --> 00:09:12,976
They'll enable soft robotic devices

198
00:09:13,000 --> 00:09:15,351
that can improve the quality of life.

199
00:09:15,375 --> 00:09:19,184
Soft robotics will enable a new generation
of more lifelike prosthetics

200
00:09:19,208 --> 00:09:21,434
for people who have lost
parts of their bodies.

201
00:09:21,458 --> 00:09:23,268
Here you see some HASELs in my lab,

202
00:09:23,292 --> 00:09:25,875
early testing,
driving a prosthetic finger.

203
00:09:27,542 --> 00:09:31,417
One day, we may even merge
our bodies with robotic parts.

204
00:09:33,125 --> 00:09:35,250
I know that sounds very scary at first.

205
00:09:36,917 --> 00:09:38,768
But when I think about my grandparents

206
00:09:38,792 --> 00:09:41,726
and the way they become
more dependent on others

207
00:09:41,750 --> 00:09:45,851
to perform simple everyday tasks
such as using the restroom alone,

208
00:09:45,875 --> 00:09:48,083
they often feel like
they're becoming a burden.

209
00:09:48,958 --> 00:09:52,226
With soft robotics, we will be able
to enhance and restore

210
00:09:52,250 --> 00:09:53,851
agility and dexterity,

211
00:09:53,875 --> 00:09:56,768
and thereby help older people
maintain autonomy

212
00:09:56,792 --> 00:09:59,059
for longer parts of their lives.

213
00:09:59,083 --> 00:10:01,583
Maybe we can call that
"robotics for antiaging"

214
00:10:03,000 --> 00:10:05,458
or even a next stage of human evolution.

215
00:10:07,083 --> 00:10:09,934
Unlike their traditional
rigid counterparts,

216
00:10:09,958 --> 00:10:14,833
soft life-like robots will safely operate
near people and help us at home.

217
00:10:15,667 --> 00:10:19,059
Soft robotics is a very young field.
We're just getting started.

218
00:10:19,083 --> 00:10:22,184
I hope that many young people
from many different backgrounds

219
00:10:22,208 --> 00:10:24,059
join us on this exciting journey

220
00:10:24,083 --> 00:10:26,226
and help shape the future of robotics

221
00:10:26,250 --> 00:10:28,833
by introducing new concepts
inspired by nature.

222
00:10:30,708 --> 00:10:32,309
If we do this right,

223
00:10:32,333 --> 00:10:34,101
we can improve the quality of life

224
00:10:34,125 --> 00:10:35,434
for all of us.

225
00:10:35,458 --> 00:10:36,726
Thank you.

226
00:10:36,750 --> 00:10:40,917
(Applause)