TEDGlobal>Geneva
Auke Ijspeert: A robot that runs and swims like a salamander
奧克．捷思皮爾: 一個出水能跑，入水能游的蜥蜴機械人
Filmed:
Readability: 4.8
2,016,271 views
機械設計學家奧克．捷思皮爾（ Auke Ijspeert）設計機械生物，就是那種可以模擬真實動物的機器人，它能夠處理複雜的場地，而且即將像科幻小說一樣出現在我們家裡。在創造這些機械生物的過程中，製造這些機器人的過程，引領出了更好的自動裝置，可以用於農耕、服務、搜索和救援。這些機械人不單只是模倣自然世界，同時亦幫助我們更認識我們的生物，解開了一些前所未知有關於脊髓的秘密。
Auke Ijspeert - Roboticist
Auke Ijspeert works at the intersection of robotics, biology and computational neuroscience. Full bio
Auke Ijspeert works at the intersection of robotics, biology and computational neuroscience. Full bio
Double-click the English transcript below to play the video.
00:12
This is Pleurobot.
0
760
1840
這是機械蠑螈，
機械蠑螈是一種遙控機具，
我們設計時，高度模擬了一類
我們設計時，高度模擬了一類
00:15
Pleurobot is a robot that we designed
to closely mimic a salamander species
to closely mimic a salamander species
1
3400
3616
00:19
called Pleurodeles waltl.
2
7040
1400
稱為歐非肋突螈的兩棲類。
你可以看到，機械蠑螈能夠走路，
00:21
Pleurobot can walk, as you can see here,
3
9240
2256
00:23
and as you'll see later, it can also swim.
4
11520
2040
稍後你也會看到，它會游泳，
你可能會問，我們為甚麼要
設計這一類的遙控機械?
設計這一類的遙控機械?
00:26
So you might ask,
why did we design this robot?
why did we design this robot?
5
14280
2191
事實上，這具機械是設計來
作為腦神經研究的科研工具。
作為腦神經研究的科研工具。
00:28
And in fact, this robot has been designed
as a scientific tool for neuroscience.
as a scientific tool for neuroscience.
6
16960
3762
我們也確實在設計的時候，
夥同腦神經學家們
夥同腦神經學家們
00:33
Indeed, we designed it
together with neurobiologists
together with neurobiologists
7
21400
2496
00:35
to understand how animals move,
8
23920
1896
去理解動物怎樣律動，
尤其去理解脊髓如何控制肢體行動。
00:37
and especially how the spinal cord
controls locomotion.
controls locomotion.
9
25840
2760
但當我在機械生物做出愈多的研究，
00:41
But the more I work in biorobotics,
10
29560
1696
00:43
the more I'm really impressed
by animal locomotion.
by animal locomotion.
11
31280
2381
我愈訝異於動物的肢體行動。
請你想像一下，一隻海豚的游泳
或一隻貓兒的奔跑跳躍，
或一隻貓兒的奔跑跳躍，
00:45
If you think of a dolphin swimming
or a cat running or jumping around,
or a cat running or jumping around,
12
33920
4296
或甚至是我們人類
00:50
or even us as humans,
13
38240
1576
在慢跑或打網球時，
00:51
when you go jogging or play tennis,
14
39840
1816
00:53
we do amazing things.
15
41680
1240
我們是在做驚奇的事情。
事實上，我們的神經系統解決了
一個非常非常複雜的控制問題。
一個非常非常複雜的控制問題。
00:55
And in fact, our nervous system solves
a very, very complex control problem.
a very, very complex control problem.
16
43880
4136
01:00
It has to coordinate
more or less 200 muscles perfectly,
more or less 200 muscles perfectly,
17
48040
3096
它需要完美得協調
大約二百組肌肉，
大約二百組肌肉，
如果協調得不好，
我們會摔倒，或行動得蹩扭。
我們會摔倒，或行動得蹩扭。
01:03
because if the coordination is bad,
we fall over or we do bad locomotion.
we fall over or we do bad locomotion.
18
51160
3680
而我的目標就是要理解這是如何運作的。
01:07
And my goal is to understand
how this works.
how this works.
19
55560
2720
動物的肢體行動，
有四個主要元件。
有四個主要元件。
01:11
There are four main components
behind animal locomotion.
behind animal locomotion.
20
59160
2840
第一個元件就是身體，
01:14
The first component is just the body,
21
62800
1936
01:16
and in fact we should never underestimate
22
64760
1976
事實上，我們都不應該低估
01:18
to what extent the biomechanics
already simplify locomotion in animals.
already simplify locomotion in animals.
23
66760
3480
生物力學已經簡化動物的
肢體行動到哪種程度。
肢體行動到哪種程度。
第二個元件就是脊髓，
01:22
Then you have the spinal cord,
24
70920
1456
01:24
and in the spinal cord you find reflexes,
25
72400
1976
你在脊髓裡可以找到反射作用，
01:26
multiple reflexes that create
a sensorimotor coordination loop
a sensorimotor coordination loop
26
74400
3456
多重反射作用會在脊髓的神經活動
與機械活動之間
與機械活動之間
01:29
between neural activity in the spinal cord
and mechanical activity.
and mechanical activity.
27
77880
3480
產生出一種
感知運動的協調迴路，
感知運動的協調迴路，
第三個元件是中樞模式產生器。
01:34
A third component
are central pattern generators.
are central pattern generators.
28
82000
2976
01:37
These are very interesting circuits
in the spinal cord of vertebrate animals
in the spinal cord of vertebrate animals
29
85000
3896
這些非常有趣的迴路，
存在於脊椎動物的脊髓裏，
存在於脊椎動物的脊髓裏，
01:40
that can generate, by themselves,
30
88920
1616
當它接收到非常簡單的輸入訊號時，
01:42
very coordinated
rhythmic patterns of activity
rhythmic patterns of activity
31
90560
2736
它們可以自己產生出
01:45
while receiving
only very simple input signals.
only very simple input signals.
32
93320
2376
非常協調且有節奏感的運動，
01:47
And these input signals
33
95720
1216
而這些輸入訊號，
01:48
coming from descending modulation
from higher parts of the brain,
from higher parts of the brain,
34
96960
3056
就是從大腦內較高部位
發射出來的「下行性調控」訊號，
發射出來的「下行性調控」訊號，
01:52
like the motor cortex,
the cerebellum, the basal ganglia,
the cerebellum, the basal ganglia,
35
100040
2696
就如同運動皮質層、
小腦，基底核一樣，
小腦，基底核一樣，
01:54
will all modulate activity
of the spinal cord
of the spinal cord
36
102760
2136
而當我們在做肢體活動的時候，
01:56
while we do locomotion.
37
104920
1456
它能夠調控脊髓的所有活動。
01:58
But what's interesting is to what extent
just a low-level component,
just a low-level component,
38
106400
3216
有趣的是，在某些程度上，
02:01
the spinal cord, together with the body,
39
109640
1936
這些脊髓，連同身體，
02:03
already solve a big part
of the locomotion problem.
of the locomotion problem.
40
111600
2456
已經可以解決大部份肢體的活動問題。
02:06
You probably know it by the fact
that you can cut the head off a chicken,
that you can cut the head off a chicken,
41
114080
3422
你大概也知道一個事實，
當一隻雞被砍頭之後，
當一隻雞被砍頭之後，
它還可以再跑一會兒，
02:09
it can still run for a while,
42
117532
1381
02:10
showing that just the lower part,
spinal cord and body,
spinal cord and body,
43
118937
2539
這表示，僅僅較低部位的脊髓和身體，
02:13
already solve a big part of locomotion.
44
121500
1873
已經解決了大部份肢體活動的問題。
02:15
Now, understanding how this works
is very complex,
is very complex,
45
123397
2459
要理解這是如何運作的，
其實也蠻複雜的，
其實也蠻複雜的，
02:17
because first of all,
46
125880
1296
因為，首先，
02:19
recording activity in the spinal cord
is very difficult.
is very difficult.
47
127200
2620
要記錄脊髓裡面的活動非常困難。
02:21
It's much easier to implant electrodes
in the motor cortex
in the motor cortex
48
129844
2772
在大腦運動皮層植入電極遠比
在脊髓植入電極容易，
在脊髓植入電極容易，
02:24
than in the spinal cord,
because it's protected by the vertebrae.
because it's protected by the vertebrae.
49
132640
3056
因為它被脊椎骨保護著。
02:27
Especially in humans, very hard to do.
50
135720
1816
尤其是在人類身上，非常難辦到。
02:29
A second difficulty is that locomotion
is really due to a very complex
is really due to a very complex
51
137560
3776
第二個困難，有很大的原因是，
肢體行動在這四個元件之間，
肢體行動在這四個元件之間，
是非常複雜且動態交互作用著的。
02:33
and very dynamic interaction
between these four components.
between these four components.
52
141360
3056
02:36
So it's very hard to find out
what's the role of each over time.
what's the role of each over time.
53
144440
3240
所以每次要找出那一個元件
擔任那一個角色，真的是很困難。
擔任那一個角色，真的是很困難。
這也是為什麼機械生物，
02:40
This is where biorobots like Pleurobot
and mathematical models
and mathematical models
54
148880
3736
02:44
can really help.
55
152640
1200
在建立像是機械蠑螈和
數學模組上很有幫助的原因。
數學模組上很有幫助的原因。
所以甚麼是機械生物呢？
02:47
So what's biorobotics?
56
155480
1256
02:48
Biorobotics is a very active field
of research in robotics
of research in robotics
57
156760
2736
機械生物是機械科研裡
一個非常活躍的領域，
一個非常活躍的領域，
02:51
where people want to
take inspiration from animals
take inspiration from animals
58
159520
2456
人們都想從動物裡得到啟發，
02:54
to make robots to go outdoors,
59
162000
2456
製成一些可以到户外去的機械人，
02:56
like service robots
or search and rescue robots
or search and rescue robots
60
164480
2656
像是一些服務業機械人，
或是可從事搜索和救援的機械人
或是可從事搜索和救援的機械人
02:59
or field robots.
61
167160
1200
或是農耕機械人。
而主要目的就是，
要從動物身上得到啟發
要從動物身上得到啟發
03:00
And the big goal here
is to take inspiration from animals
is to take inspiration from animals
62
168880
2696
03:03
to make robots that can handle
complex terrain --
complex terrain --
63
171600
2336
來製造一些機械人，
可以處理一些複雜的地形--
可以處理一些複雜的地形--
03:05
stairs, mountains, forests,
64
173960
1616
像是樓梯、山脈、森林、
03:07
places where robots
still have difficulties
still have difficulties
65
175600
2016
一些機械人仍然遇到困難的地方，
03:09
and where animals
can do a much better job.
can do a much better job.
66
177640
2056
以及動物可以做得更好的地方。
03:11
The robot can be a wonderful
scientific tool as well.
scientific tool as well.
67
179720
2496
機械人同樣也是神奇的科研工具，
03:14
There are some very nice projects
where robots are used,
where robots are used,
68
182240
2620
有些很棒的科研項目利用機械人
03:16
like a scientific tool for neuroscience,
for biomechanics or for hydrodynamics.
for biomechanics or for hydrodynamics.
69
184884
3972
做為腦神經、生物力學
或水力學的科研工具。
或水力學的科研工具。
03:20
And this is exactly
the purpose of Pleurobot.
the purpose of Pleurobot.
70
188880
2120
而這就是做機械蠑螈的目的。
在我的實驗室，我們夥同腦神經生物學家
03:23
So what we do in my lab
is to collaborate with neurobiologists
is to collaborate with neurobiologists
71
191600
2936
03:26
like Jean-Marie Cabelguen,
a neurobiologist in Bordeaux in France,
a neurobiologist in Bordeaux in France,
72
194560
3216
例如法國波爾多的腦神經生物學家
Jean-Marie Cabelguen，
Jean-Marie Cabelguen，
03:29
and we want to make spinal cord models
and validate them on robots.
and validate them on robots.
73
197800
4040
我們打算製作出脊髓的模型，
然後在機器人上驗証。
然後在機器人上驗証。
我們希望從簡單出發。
03:34
And here we want to start simple.
74
202480
1616
03:36
So it's good to start with simple animals
75
204120
1976
所以從簡單的動物開始就好，
03:38
like lampreys, which are
very primitive fish,
very primitive fish,
76
206120
2256
像是七鰓鰻，非常原始的魚類，
03:40
and then gradually
go toward more complex locomotion,
go toward more complex locomotion,
77
208400
2496
然後逐漸地邁向更複雜的肢體活動，
03:42
like in salamanders,
78
210920
1256
像是蜥蜴，
03:44
but also in cats and in humans,
79
212200
1496
但也包含貓、人類，
03:45
in mammals.
80
213720
1200
哺乳動物等。
03:47
And here, a robot becomes
an interesting tool
an interesting tool
81
215880
2376
所以，機械人成為了一個
03:50
to validate our models.
82
218280
1936
可以驗証我們模型的有趣工具。
03:52
And in fact, for me, Pleurobot
is a kind of dream becoming true.
is a kind of dream becoming true.
83
220240
3016
事實上對我來說，機械蠑螈
算是圓了我一個夢想。
算是圓了我一個夢想。
03:55
Like, more or less 20 years ago
I was already working on a computer
I was already working on a computer
84
223280
3256
大概二十年前，在我博士班的期間，
03:58
making simulations of lamprey
and salamander locomotion
and salamander locomotion
85
226560
2656
我已經在電腦上，製作一些
04:01
during my PhD.
86
229240
1536
七鰓鰻和蜥蜴肢體活動的模擬，
04:02
But I always knew that my simulations
were just approximations.
were just approximations.
87
230800
3376
但我一直以來都知道，
我的模擬只是粗略概算。
我的模擬只是粗略概算。
04:06
Like, simulating the physics in water
or with mud or with complex ground,
or with mud or with complex ground,
88
234200
3976
像是在水中模擬物理現象，
或是在混雜泥土裡或是複雜的地表面上，
或是在混雜泥土裡或是複雜的地表面上，
04:10
it's very hard to simulate that
properly on a computer.
properly on a computer.
89
238200
2656
這些都是很難在電腦上適當地模擬出來的。
04:12
Why not have a real robot
and real physics?
and real physics?
90
240880
2040
為什麼不乾脆做一個
真實的機械人或真實的物體？
真實的機械人或真實的物體？
04:15
So among all these animals,
one of my favorites is the salamander.
one of my favorites is the salamander.
91
243600
3136
在眾多的動物裡，蜥蜴是我喜歡的其中之一。
04:18
You might ask why,
and it's because as an amphibian,
and it's because as an amphibian,
92
246760
3456
你大概想知道為什麼，
因為以兩棲動物而言，
因為以兩棲動物而言，
04:22
it's a really key animal
from an evolutionary point of view.
from an evolutionary point of view.
93
250240
2856
從進化的角度來看，
蜥蜴其實是很重要的動物。
蜥蜴其實是很重要的動物。
04:25
It makes a wonderful link
between swimming,
between swimming,
94
253120
2056
它完美的串聯起
04:27
as you find it in eels or fish,
95
255200
1896
水棲動物的游泳
（像是鰻魚或魚）
（像是鰻魚或魚）
04:29
and quadruped locomotion,
as you see in mammals, in cats and humans.
as you see in mammals, in cats and humans.
96
257120
4120
以及哺乳類動物的四肢活動
（像是貓或人）。
（像是貓或人）。
事實上，現代的蜥蜴
04:34
And in fact, the modern salamander
97
262160
1656
04:35
is very close to the first
terrestrial vertebrate,
terrestrial vertebrate,
98
263840
2376
與第一代的陸棲脊椎動物非常相近，
04:38
so it's almost a living fossil,
99
266240
1536
幾乎就是一種活化石，
04:39
which gives us access to our ancestor,
100
267800
1936
讓我們可以接近自己的祖宗，
04:41
the ancestor to all terrestrial tetrapods.
101
269760
2920
所有陸棲四肢動物的祖宗。
蜥蜴是藉由一種稱為鰻游的泳態，
04:45
So the salamander swims
102
273240
1376
04:46
by doing what's called
an anguilliform swimming gait,
an anguilliform swimming gait,
103
274640
2496
來進行游泳的動作，
04:49
so they propagate a nice traveling wave
of muscle activity from head to tail.
of muscle activity from head to tail.
104
277160
3640
它們從頭部到尾部的肌肉活動，
傳遞出一種很優美的游行波浪。
傳遞出一種很優美的游行波浪。
而當你把蜥蜴放在地面上時，
04:53
And if you place
the salamander on the ground,
the salamander on the ground,
105
281440
2176
04:55
it switches to what's called
a walking trot gait.
a walking trot gait.
106
283640
2336
它又會轉化為快走的步態。
04:58
In this case, you have nice
periodic activation of the limbs
periodic activation of the limbs
107
286000
2863
在這個案例，你有很好的
週期性肢體律動
週期性肢體律動
05:00
which are very nicely coordinated
108
288887
1609
可以非常好地協調出
05:02
with this standing wave
undulation of the body,
undulation of the body,
109
290520
2656
這樣持續性波浪的身體起伏，
05:05
and that's exactly the gait
that you are seeing here on Pleurobot.
that you are seeing here on Pleurobot.
110
293200
3656
就如你們現在所看到的
機械蠑螈的步態。
機械蠑螈的步態。
05:08
Now, one thing which is very surprising
and fascinating in fact
and fascinating in fact
111
296880
2976
事實上，其中一件很令人訝異
卻又讚嘆的事實就是...
卻又讚嘆的事實就是...
05:11
is the fact that all this can be generated
just by the spinal cord and the body.
just by the spinal cord and the body.
112
299880
4136
這些活動可以僅藉由
脊髓和身體就可以啟動了。
脊髓和身體就可以啟動了。
05:16
So if you take
a decerebrated salamander --
a decerebrated salamander --
113
304040
2000
所以即使是一隻沒有腦袋的蜥蜴 --
05:18
it's not so nice
but you remove the head --
but you remove the head --
114
306064
2016
那不是太好, 但當移除了頭顱--
05:20
and if you electrically
stimulate the spinal cord,
stimulate the spinal cord,
115
308104
2672
而你用電殛刺激脊髓，
05:22
at low level of stimulation
this will induce a walking-like gait.
this will induce a walking-like gait.
116
310800
3256
在低電流的刺激下，
會做出走路一樣的步態。
會做出走路一樣的步態。
05:26
If you stimulate a bit more,
the gait accelerates.
the gait accelerates.
117
314080
2456
如果你稍稍加強刺激度，
步伐就會隨之加快。
步伐就會隨之加快。
05:28
And at some point, there's a threshold,
118
316560
1896
到了若干程度，會有一個臨界點，
05:30
and automatically,
the animal switches to swimming.
the animal switches to swimming.
119
318480
2536
隨後，動物會自動地
從行走轉為游泳
從行走轉為游泳
05:33
This is amazing.
120
321040
1376
這真是神乎其技。
05:34
Just changing the global drive,
121
322440
1496
只是改變了中央的驅動器，
05:35
as if you are pressing the gas pedal
122
323960
1736
就如同你在踩油門一樣，
05:37
of descending modulation
to your spinal cord,
to your spinal cord,
123
325720
2136
把下行性調控訊號傳遞到你的脊髓，
05:39
makes a complete switch
between two very different gaits.
between two very different gaits.
124
327880
3000
在兩種不一樣的模式間相互切換。
05:44
And in fact, the same
has been observed in cats.
has been observed in cats.
125
332440
2576
其實同樣的情況，
在貓身上也觀察得到，
在貓身上也觀察得到，
05:47
If you stimulate the spinal cord of a cat,
126
335040
2016
如果你刺激一隻貓的脊髓，
05:49
you can switch between
walk, trot and gallop.
walk, trot and gallop.
127
337080
2216
你能夠在其間切換模式：
行走、緩跑和急步跑。
行走、緩跑和急步跑。
05:51
Or in birds, you can make a bird
switch between walking,
switch between walking,
128
339320
2736
或在鳥類身上，你可以隨興
切換一隻小鳥，
切換一隻小鳥，
05:54
at a low level of stimulation,
129
342080
1456
在低電流時，走路，
05:55
and flapping its wings
at high-level stimulation.
at high-level stimulation.
130
343560
2816
在高電流刺激時，揮動翅膀。
05:58
And this really shows that the spinal cord
131
346400
2016
而這告訴我們，
脊髓是個非常複雜精密的
肢體行動控制器。
肢體行動控制器。
06:00
is a very sophisticated
locomotion controller.
locomotion controller.
132
348440
2416
06:02
So we studied salamander locomotion
in more detail,
in more detail,
133
350880
2456
於是我們更仔細的研究蜥蜴的肢體行動，
06:05
and we had in fact access
to a very nice X-ray video machine
to a very nice X-ray video machine
134
353360
3096
其我們有一部很好的X光錄影機，
06:08
from Professor Martin Fischer
in Jena University in Germany.
in Jena University in Germany.
135
356480
3576
是由德國 Jena 大學的 Martin Fischer 教授所提供。
06:12
And thanks to that,
you really have an amazing machine
you really have an amazing machine
136
360080
2576
感謝有這部神奇的機器，
06:14
to record all the bone motion
in great detail.
in great detail.
137
362680
2456
把所有的骨骼行動都仔細的紀錄下來。
06:17
That's what we did.
138
365160
1256
這就是我們在做的事。
06:18
So we basically figured out
which bones are important for us
which bones are important for us
139
366440
3176
基本上，我們找出了
對我們來說重要的骨骼，
對我們來說重要的骨骼，
06:21
and collected their motion in 3D.
140
369640
3016
並且收集它們的3D動作。
06:24
And what we did is collect
a whole database of motions,
a whole database of motions,
141
372680
2696
我們所做的就是收集
整個骨骼的動作資料庫，
整個骨骼的動作資料庫，
06:27
both on ground and in water,
142
375400
1656
從水上到陸上，
06:29
to really collect a whole database
of motor behaviors
of motor behaviors
143
377080
2484
實際地去收集一隻動物所有的
06:31
that a real animal can do.
144
379589
1244
移動行為資料庫。
06:32
And then our job as roboticists
was to replicate that in our robot.
was to replicate that in our robot.
145
380858
3150
而我們機械設計學家的工作就是，
將這些資料複製到我們的機械人。
將這些資料複製到我們的機械人。
06:36
So we did a whole optimization process
to find out the right structure,
to find out the right structure,
146
384033
3383
所以我們做了全方位的優化程序
來找出正確的結構、
來找出正確的結構、
06:39
where to place the motors,
how to connect them together,
how to connect them together,
147
387440
2656
在哪裡放置馬達、
06:42
to be able to replay
these motions as well as possible.
these motions as well as possible.
148
390120
2880
如何把它們連接一起，
盡可能地重製出這些動作等等。
盡可能地重製出這些動作等等。
06:45
And this is how Pleurobot came to life.
149
393680
2360
機械蠑螈就是這樣成型的。
讓我們來看看它跟
真正的動物有多近似。
真正的動物有多近似。
06:49
So let's look at how close
it is to the real animal.
it is to the real animal.
150
397200
2416
你現在看到的是，
真正的動物和機械蠑螈在行走時
真正的動物和機械蠑螈在行走時
06:52
So what you see here
is almost a direct comparison
is almost a direct comparison
151
400960
2496
06:55
between the walking
of the real animal and the Pleurobot.
of the real animal and the Pleurobot.
152
403480
2696
直接對比的影片。
06:58
You can see that we have
almost a one-to-one exact replay
almost a one-to-one exact replay
153
406200
2736
你可以看到幾乎是一比一的比例，
07:00
of the walking gait.
154
408960
1256
重演着走路的步態。
07:02
If you go backwards and slowly,
you see it even better.
you see it even better.
155
410240
2600
如果你倒退或慢動作，你可以看得更清楚。
更棒的是，我們可以游泳。
07:07
But even better, we can do swimming.
156
415520
2376
07:09
So for that we have a dry suit
that we put all over the robot --
that we put all over the robot --
157
417920
3016
我們甚至為機械蠑螈穿上了潛水衣--
07:12
(Laughter)
158
420960
1096
（笑聲）
07:14
and then we can go in water
and start replaying the swimming gaits.
and start replaying the swimming gaits.
159
422080
3176
然後我們可以到水裡，
開始重製游泳的泳態。
開始重製游泳的泳態。
07:17
And here, we were very happy,
because this is difficult to do.
because this is difficult to do.
160
425280
3336
我們對於此很高興，
因為這個真的很難。
因為這個真的很難。
07:20
The physics of interaction are complex.
161
428640
2216
互動的物理現象相當複雜。
07:22
Our robot is much bigger
than a small animal,
than a small animal,
162
430880
2416
我們的機械蠑螈要比小動物大很多，
07:25
so we had to do what's called
dynamic scaling of the frequencies
dynamic scaling of the frequencies
163
433320
3056
所以我們得找出
稱之為「等比例動態」的頻率，
稱之為「等比例動態」的頻率，
07:28
to make sure we had
the same interaction physics.
the same interaction physics.
164
436400
2336
來確定我們也得到了
一樣的互動物理現象。
一樣的互動物理現象。
07:30
But you see at the end,
we have a very close match,
we have a very close match,
165
438760
2416
你可以看到，
最後我們可以非常接近地運動，
最後我們可以非常接近地運動，
07:33
and we were very, very happy with this.
166
441200
1880
所以我們對此非常非常的高興。
07:35
So let's go to the spinal cord.
167
443480
2216
現在我們來看看脊髓。
07:37
So here what we did
with Jean-Marie Cabelguen
with Jean-Marie Cabelguen
168
445720
2296
我們跟 Jean-Marie Cabelguen 一起
07:40
is model the spinal cord circuits.
169
448040
2240
模擬了脊髓的迴路。
有趣的是，蜥蜴
07:43
And what's interesting
is that the salamander
is that the salamander
170
451040
2136
07:45
has kept a very primitive circuit,
171
453200
1620
保持了最原始的迴路，
07:46
which is very similar
to the one we find in the lamprey,
to the one we find in the lamprey,
172
454844
2652
非常相近於我們找到的七鰓鰻，
07:49
this primitive eel-like fish,
173
457520
1976
這個像鰻魚的原始魚類，
07:51
and it looks like during evolution,
174
459520
1736
看起來像是在進化期間，
07:53
new neural oscillators
have been added to control the limbs,
have been added to control the limbs,
175
461280
2936
有新的神經振動器
會被加進來去控制肢體
會被加進來去控制肢體
07:56
to do the leg locomotion.
176
464240
1416
來帶動腿的行動。
07:57
And we know where
these neural oscillators are
these neural oscillators are
177
465680
2176
我們知道這些神經振動器在哪裡，
07:59
but what we did was to make
a mathematical model
a mathematical model
178
467880
2256
但我們要做的是，計算出數學模式，
08:02
to see how they should be coupled
179
470160
1616
看看怎樣把他們配對起來，
08:03
to allow this transition
between the two very different gaits.
between the two very different gaits.
180
471800
2936
來讓這兩種非常不同的
動作可以自由轉換。
動作可以自由轉換。
08:06
And we tested that on board of a robot.
181
474760
2560
我們就在機械蠑螈的電板上測試。
08:09
And this is how it looks.
182
477680
1200
而它看起來就像是這樣。
08:18
So what you see here
is a previous version of Pleurobot
is a previous version of Pleurobot
183
486920
3016
這裡你們看到的是，
上一代版本的機械蠑螈，
上一代版本的機械蠑螈，
08:21
that's completely controlled
by our spinal cord model
by our spinal cord model
184
489960
3096
完全由我們輸入在電路板上
08:25
programmed on board of the robot.
185
493080
1600
的脊髓模組程式所控制。
08:27
And the only thing we do
186
495280
1216
我們唯一做的是，
08:28
is send to the robot
through a remote control
through a remote control
187
496520
2176
透過遥控器，向機械人發出
08:30
the two descending signals
it normally should receive
it normally should receive
188
498720
2496
兩組下行性調控訊號，而這通常源自於
08:33
from the upper part of the brain.
189
501240
1600
腦部的上半部分。
08:35
And what's interesting is,
by playing with these signals,
by playing with these signals,
190
503480
2696
有趣的是，通過這些訊號
我們可以完全控制速度、前進、步、泳態。
08:38
we can completely control
speed, heading and type of gait.
speed, heading and type of gait.
191
506200
2800
08:41
For instance,
192
509600
1216
比方說，
08:42
when we stimulate at a low level,
we have the walking gait,
we have the walking gait,
193
510840
3576
當我們透過低電流作出刺激時，
我們得到的是行走的狀態，
我們得到的是行走的狀態，
08:46
and at some point, if we stimulate a lot,
194
514440
1976
來到某種程度，
如果我們加強了刺激 ，
如果我們加強了刺激 ，
08:48
very rapidly it switches
to the swimming gait.
to the swimming gait.
195
516440
2160
它會迅速地轉化為游泳的狀態。
08:51
And finally, we can also
do turning very nicely
do turning very nicely
196
519480
2216
最後，我們也可輕鬆的轉向
08:53
by just stimulating more one side
of the spinal cord than the other.
of the spinal cord than the other.
197
521720
3520
主要在脊髓左右兩邊，
在其中的一邊加以刺激就可以了。
在其中的一邊加以刺激就可以了。
我覺得這真是漂亮
08:58
And I think it's really beautiful
198
526200
1616
08:59
how nature has distributed control
199
527840
2256
自然界先天的分配了控制權
09:02
to really give a lot of responsibility
to the spinal cord
to the spinal cord
200
530120
2856
把很多責任交付予脊髓，
09:05
so that the upper part of the brain
doesn't need to worry about every muscle.
doesn't need to worry about every muscle.
201
533000
3656
所以大腦的上半部分
不需要再煩惱每一條肌肉。
不需要再煩惱每一條肌肉。
09:08
It just has to worry
about this high-level modulation,
about this high-level modulation,
202
536680
2536
大腦只負擔高層次的調節，
09:11
and it's really the job of the spinal cord
to coordinate all the muscles.
to coordinate all the muscles.
203
539240
3576
協調各肌肉的任務，
就交付予脊髓了。
就交付予脊髓了。
09:14
So now let's go to cat locomotion
and the importance of biomechanics.
and the importance of biomechanics.
204
542840
3520
現在我們來看看貓的行動
和生物力學的重要性。
和生物力學的重要性。
09:19
So this is another project
205
547080
1256
這是另一個項目，
09:20
where we studied cat biomechanics,
206
548360
2416
我們研究貓的生物力學，
09:22
and we wanted to see how much
the morphology helps locomotion.
the morphology helps locomotion.
207
550800
3896
而我們想知道形態學
對於肢體活動的幫助。
對於肢體活動的幫助。
09:26
And we found three important
criteria in the properties,
criteria in the properties,
208
554720
3616
我們得出了三個性質的標準，
09:30
basically, of the limbs.
209
558360
1320
基本上，就是肢體內的性質。
09:32
The first one is that a cat limb
210
560320
1976
首先就是貓的肢體，
09:34
more or less looks
like a pantograph-like structure.
like a pantograph-like structure.
211
562320
2696
大概類似導電弓架的結構。
09:37
So a pantograph is a mechanical structure
212
565040
2216
導電弓架是一個機電的結構
09:39
which keeps the upper segment
and the lower segments always parallel.
and the lower segments always parallel.
213
567280
3400
永恆的保持着
上部份和下部份的平行。
上部份和下部份的平行。
09:43
So a simple geometrical system
that kind of coordinates a bit
that kind of coordinates a bit
214
571600
3096
其實就是一個簡單的幾何系統，
09:46
the internal movement of the segments.
215
574720
1816
協調着各部位的內部移動。
09:48
A second property of cat limbs
is that they are very lightweight.
is that they are very lightweight.
216
576560
3056
貓兒肢體的第二個性質是非常輕量。
09:51
Most of the muscles are in the trunk,
217
579640
1856
大部份的肌肉集中在驅體內，
09:53
which is a good idea,
because then the limbs have low inertia
because then the limbs have low inertia
218
581520
2896
這是很棒的點子，因為這樣
肢體不會有低度的惰性
肢體不會有低度的惰性
09:56
and can be moved very rapidly.
219
584440
1776
反而能夠迅速的活動。
09:58
The last final important property is this
very elastic behavior of the cat limb,
very elastic behavior of the cat limb,
220
586240
3816
最後很重要的性質是，貓的肢體彈力很強，
10:02
so to handle impacts and forces.
221
590080
2656
有利於處理好衝擊力和震盪力。
10:04
And this is how we designed Cheetah-Cub.
222
592760
2336
我們也是如此設計小獵豹的。
10:07
So let's invite Cheetah-Cub onstage.
223
595120
2200
現在有請小獵豹到台上來。
10:14
So this is Peter Eckert,
who does his PhD on this robot,
who does his PhD on this robot,
224
602160
3656
這位是 Peter Eckert，
他用這部機械人作他的博士學位研究，
他用這部機械人作他的博士學位研究，
10:17
and as you see, it's a cute little robot.
225
605840
2056
你可以看到，這是一隻可愛的小機械。
10:19
It looks a bit like a toy,
226
607920
1256
它看起來有點像是玩具，
10:21
but it was really used
as a scientific tool
as a scientific tool
227
609200
2056
但其實卻是個科研工具，
10:23
to investigate these properties
of the legs of the cat.
of the legs of the cat.
228
611280
3296
用來查證貓兒四腿的特質。
10:26
So you see, it's very compliant,
very lightweight,
very lightweight,
229
614600
2616
你可以看到，它非常柔韌，非常輕量，
10:29
and also very elastic,
230
617240
1256
同時也非常的彈性，
10:30
so you can easily press it down
and it will not break.
and it will not break.
231
618520
2776
所以你可以很輕鬆的
把它壓下而絲毫不損。
把它壓下而絲毫不損。
10:33
It will just jump, in fact.
232
621320
1456
實際上，它只會彈跳。
10:34
And this very elastic property
is also very important.
is also very important.
233
622800
2880
這個彈力的性質也非常重要。
你同時也可以看到一些其他特質，
10:39
And you also see a bit these properties
234
627160
1896
10:41
of these three segments
of the leg as pantograph.
of the leg as pantograph.
235
629080
2400
就是導電弓架作為腳部的三個部份。
10:44
Now, what's interesting
is that this quite dynamic gait
is that this quite dynamic gait
236
632280
2776
有趣的是，這很有動態，
10:47
is obtained purely in open loop,
237
635080
1896
純粹來自開放式的迴路，
10:49
meaning no sensors,
no complex feedback loops.
no complex feedback loops.
238
637000
3136
意思就是沒有探測器，
也沒有複雜的回饋迴路。
也沒有複雜的回饋迴路。
10:52
And that's interesting, because it means
239
640160
2416
那是很有趣的，因為那意味著，
10:54
that just the mechanics
already stabilized this quite rapid gait,
already stabilized this quite rapid gait,
240
642600
4016
單靠機械模式,
已能把這快速的步態穩定下來，
已能把這快速的步態穩定下來，
10:58
and that really good mechanics
already basically simplify locomotion.
already basically simplify locomotion.
241
646640
4176
所以只需要有很好的機械性，
基本上已能夠簡化肢體的活動。
基本上已能夠簡化肢體的活動。
11:02
To the extent that we can even
disturb a bit locomotion,
disturb a bit locomotion,
242
650840
3296
在某程度上, 我們甚至可以對肢體行動，
作出一些干預，
作出一些干預，
11:06
as you will see in the next video,
243
654160
1656
就正如你將看到的影像一樣，
11:07
where we can for instance do some exercise
where we have the robot go down a step,
where we have the robot go down a step,
244
655840
3896
比方說我們可以做一些運動，
讓機械人下樓梯，
讓機械人下樓梯，
11:11
and the robot will not fall over,
245
659760
1616
而且機械人不會跌倒，
11:13
which was a surprise for us.
246
661400
1576
那其實讓我們很訝異。
11:15
This is a small perturbation.
247
663000
1416
這是輕微的干擾。
11:16
I was expecting the robot
to immediately fall over,
to immediately fall over,
248
664440
2416
我預期機械會立刻翻倒，
11:18
because there are no sensors,
no fast feedback loop.
no fast feedback loop.
249
666880
2436
因為沒有探測器，也沒有迅速回饋的迴路。
11:21
But no, just the mechanics
stabilized the gait,
stabilized the gait,
250
669340
2196
但不，單靠機械模式已能把這步態穩定下來，
11:23
and the robot doesn't fall over.
251
671560
1576
而機械人並沒有翻倒。
11:25
Obviously, if you make the step bigger,
and if you have obstacles,
and if you have obstacles,
252
673160
3136
顯而易見，如果你跨大了步伐，
再加上你遇到了障礙物，
再加上你遇到了障礙物，
11:28
you need the full control loops
and reflexes and everything.
and reflexes and everything.
253
676320
3656
你就會需要完整的控制迴路，
還要所有的反射動作及每樣東西。
還要所有的反射動作及每樣東西。
11:32
But what's important here
is that just for small perturbation,
is that just for small perturbation,
254
680000
2936
這裡重要的是，只是輕微的干擾，
11:34
the mechanics are right.
255
682960
1496
機械性能就正確了。
11:36
And I think this is
a very important message
a very important message
256
684480
2096
我相信這是很重要的訊息，
11:38
from biomechanics and robotics
to neuroscience,
to neuroscience,
257
686600
2191
從生物力學及機械學以至腦神經科學，
11:40
saying don't underestimate to what extent
the body already helps locomotion.
the body already helps locomotion.
258
688815
4680
不要低估身體其實在某種程度，
已經幫助了肢體行動。
已經幫助了肢體行動。
11:47
Now, how does this relate
to human locomotion?
to human locomotion?
259
695440
2160
現在，這個跟人類肢體活動有甚麼關係呢？
11:49
Clearly, human locomotion is more complex
than cat and salamander locomotion,
than cat and salamander locomotion,
260
697960
3640
非常明顯，人類的肢體活動，
遠遠比貓和蜥蜴的複雜，
遠遠比貓和蜥蜴的複雜，
11:54
but at the same time, the nervous system
of humans is very similar
of humans is very similar
261
702360
3136
但同時間，人類的神經系統
11:57
to that of other vertebrates.
262
705520
1576
卻又和其他的脊椎動物非常類近。
11:59
And especially the spinal cord
263
707120
1456
尤其是脊髓，
12:00
is also the key controller
for locomotion in humans.
for locomotion in humans.
264
708600
2640
同樣也是人類肢體行動的關鍵控制器。
所以，這也是為什麼當脊髓有損害時，
12:03
That's why, if there's a lesion
of the spinal cord,
of the spinal cord,
265
711760
2416
12:06
this has dramatic effects.
266
714200
1496
會帶來很嚴重的後果。
12:07
The person can become
paraplegic or tetraplegic.
paraplegic or tetraplegic.
267
715720
2776
那個人可能會變成
下半身癱瘓或四肢癱瘓。
下半身癱瘓或四肢癱瘓。
12:10
This is because the brain
loses this communication
loses this communication
268
718520
2376
這是因為大腦失去了
12:12
with the spinal cord.
269
720920
1256
與脊髓的溝通。
12:14
Especially, it loses
this descending modulation
this descending modulation
270
722200
2216
尤其是大腦失去了啟動
12:16
to initiate and modulate locomotion.
271
724440
1920
和協調肢體活動作用的下行性調控訊號時。
12:19
So a big goal of neuroprosthetics
272
727640
1696
所以神經義肢的一個很大的目標
12:21
is to be able to reactivate
that communication
that communication
273
729360
2376
就是可以重啟這個溝通，
12:23
using electrical or chemical stimulations.
274
731760
2440
借助電子或化學的刺激。
12:26
And there are several teams
in the world that do exactly that,
in the world that do exactly that,
275
734840
2936
世界上有好幾個隊伍
正在進行這樣的項目，
正在進行這樣的項目，
12:29
especially at EPFL.
276
737800
1216
尤其是 EPFL 的隊伍。
12:31
My colleagues Grégoire Courtine
and Silvestro Micera,
and Silvestro Micera,
277
739040
2496
和我一起合作的同事
12:33
with whom I collaborate.
278
741560
1240
Grégoire Courtine 和 Silvestro Micera 。
12:35
But to do this properly,
it's very important to understand
it's very important to understand
279
743960
3096
要恰當地做好這事情，務必要明白
12:39
how the spinal cord works,
280
747080
1736
脊髓如何運作，
12:40
how it interacts with the body,
281
748840
1696
它怎樣和身體互動，
12:42
and how the brain
communicates with the spinal cord.
communicates with the spinal cord.
282
750560
2480
而大腦又怎樣和脊髓溝通。
12:45
This is where the robots
and models that I've presented today
and models that I've presented today
283
753800
2896
今天我向各位展示的機械和模型
12:48
will hopefully play a key role
284
756720
1896
希望就是在這些非常重要的目標上
12:50
towards these very important goals.
285
758640
2656
可以做出貢獻。
12:53
Thank you.
286
761320
1216
謝謝。
12:54
(Applause)
287
762560
4560
(掌聲)
13:04
Bruno Giussani: Auke, I've seen
in your lab other robots
in your lab other robots
288
772100
2636
Bruno Giussani: 奧克, 我曾經在
你的實驗室裡，看過其他的機械人，
你的實驗室裡，看過其他的機械人，
13:06
that do things like swim in pollution
289
774760
2456
做一些例如在污染環境中游泳的事情，
13:09
and measure the pollution while they swim.
290
777240
2456
並在游泳期間測量污染程度。
13:11
But for this one,
291
779720
1216
但這一個，
13:12
you mentioned in your talk,
like a side project,
like a side project,
292
780960
3480
你在演講時表示，這只是一個子項目，
負責搜索和救援，
13:17
search and rescue,
293
785640
1216
它在鼻子上也安裝了照相機。
13:18
and it does have a camera on its nose.
294
786880
2176
13:21
Auke Ijspeert: Absolutely. So the robot --
295
789080
2496
奧克．捷思皮爾：對的。所以這個機械人--
13:23
We have some spin-off projects
296
791600
1429
我們有一些是分拆開來的項目
13:25
where we would like to use the robots
to do search and rescue inspection,
to do search and rescue inspection,
297
793053
3443
我們希望主要用作
搜索和救援的調察，
搜索和救援的調察，
13:28
so this robot is now seeing you.
298
796520
1576
所以這位機械人現正看着你。
13:30
And the big dream is to,
if you have a difficult situation
if you have a difficult situation
299
798120
3176
而遠大的夢想就是，一旦你遭遇到困難
13:33
like a collapsed building
or a building that is flooded,
or a building that is flooded,
300
801320
3616
例如倒塌的建築物，
或是被淹沒的建築物，
或是被淹沒的建築物，
這樣的情況，對救援隊、甚至救援狗而言
都是非常危險的，
都是非常危險的，
13:36
and this is very dangerous
for a rescue team or even rescue dogs,
for a rescue team or even rescue dogs,
301
804960
3336
13:40
why not send in a robot
that can crawl around, swim, walk,
that can crawl around, swim, walk,
302
808320
2896
為什麼不用可以四處爬行、游泳
和行走的機械人取替，
和行走的機械人取替，
13:43
with a camera onboard
to do inspection and identify survivors
to do inspection and identify survivors
303
811240
3176
並加上照相機來協助調察和找出生還者，
甚至有可能與生還者建立溝通。
13:46
and possibly create
a communication link with the survivor.
a communication link with the survivor.
304
814440
2776
BG: 當然，假設生還者沒有被它的外型嚇到。
13:49
BG: Of course, assuming the survivors
don't get scared by the shape of this.
don't get scared by the shape of this.
305
817240
3576
13:52
AI: Yeah, we should probably
change the appearance quite a bit,
change the appearance quite a bit,
306
820840
3296
AI: 對，我們大概也需要
把它的外型改良一下，
把它的外型改良一下，
不然生還者
可能會心臟病發而死亡，
可能會心臟病發而死亡，
13:56
because here I guess a survivor
might die of a heart attack
might die of a heart attack
307
824160
2816
13:59
just of being worried
that this would feed on you.
that this would feed on you.
308
827000
2536
僅是擔心它會把你吃掉。
14:01
But by changing the appearance
and it making it more robust,
and it making it more robust,
309
829560
2856
把它的外型改良一下，
同時令它更耐用，
同時令它更耐用，
14:04
I'm sure we can make
a good tool out of it.
a good tool out of it.
310
832440
2056
我深信這能夠成為很好的工具。
14:06
BG: Thank you very much.
Thank you and your team.
Thank you and your team.
311
834520
2286
非常感謝。謝謝你和你的團隊。
ABOUT THE SPEAKER
Auke Ijspeert - RoboticistAuke Ijspeert works at the intersection of robotics, biology and computational neuroscience.
Why you should listen
Auke Ijspeert is a professor at the EPFL (the Swiss Federal Institute of Technology at Lausanne), and head of the Biorobotics Laboratory (BioRob). He has a BSc/MSc in Physics from the EPFL and a PhD in artificial intelligence from the University of Edinburgh, with John Hallam and David Willshaw as advisors. He carried out postdocs at IDSIA and EPFL with Jean-Daniel Nicoud and Luca Gambardella, and at the University of Southern California, with Michael Arbib and Stefan Schaal.
Ijspeert is interested in using numerical simulations and robots to get a better understanding of animal locomotion and movement control, and in using inspiration from biology to design novel types of robots and locomotion controllers.
Auke Ijspeert | Speaker | TED.com