A robot that runs and swims like a salamander | Auke Ijspeert

翻译人员: Sameul George 校对人员: Jiawei Ni

This is Pleurobot.

这是一只蝾螈机器人

Pleurobot is a robot that we designed to closely mimic a salamander species

是一种我们根据蝾螈设计出的仿生机器人

called Pleurodeles waltl.

我们把它称为蝾螈机器人

Pleurobot can walk, as you can see here,

就想你看到的，它会爬行

and as you'll see later, it can also swim.

一会儿你还会看到它游泳

So you might ask, why did we design this robot?

你可能会问我们为什么设计这样一个机器人

And in fact, this robot has been designed as a scientific tool for neuroscience.

其实，蝾螈机器人是一种用来研究神经科学的工具

Indeed, we designed it together with neurobiologists

我们和神经科学家一起设计出蝾螈机器人

to understand how animals move,

希望能更好理解动物的运动原理

and especially how the spinal cord controls locomotion.

尤其是脊髓控制运动的机理

But the more I work in biorobotics,

但是，我们对仿生机器人的研究越深入

the more I'm really impressed by animal locomotion.

就越被动物运动的复杂原理所震撼

If you think of a dolphin swimming or a cat running or jumping around,

试想一下在海里遨游的海豚、又跑又跳的小猫

or even us as humans,

甚至还有我们人类

when you go jogging or play tennis,

在进行慢跑或者打网球的时候

we do amazing things.

我们的每一个动作都那么神奇

And in fact, our nervous system solves a very, very complex control problem.

实际上，在运动时，神经系统的控制过程十分复杂

It has to coordinate more or less 200 muscles perfectly,

神经系统要协调200块左右的肌肉一起运作

because if the coordination is bad, we fall over or we do bad locomotion.

一旦协调紊乱，我们就会摔跤或者肢体不协调

And my goal is to understand how this works.

而我的目标就是要弄清楚这种协调机制的原理

There are four main components behind animal locomotion.

动物的运动需要四个部分

The first component is just the body,

首先就是肢体

and in fact we should never underestimate

生物力学对于动物肢体运动的简化

to what extent the biomechanics already simplify locomotion in animals.

已经到了令人难以想象地步

Then you have the spinal cord,

第二个部分就是脊髓

and in the spinal cord you find reflexes,

脊髓中有很多反射神经

multiple reflexes that create a sensorimotor coordination loop

这些反射神经在脊髓中的神经活动和机械活动之间

between neural activity in the spinal cord and mechanical activity.

建立了感觉协调环

A third component are central pattern generators.

第三个要素是中枢模式发生器

These are very interesting circuits in the spinal cord of vertebrate animals

脊椎动物的脊髓中有一些有趣的电路

that can generate, by themselves,

这些电路在接收简单的信号时

very coordinated rhythmic patterns of activity

可以自发地产生非常协调的

while receiving only very simple input signals.

节律运动模式

And these input signals

在我们运动的时候

coming from descending modulation from higher parts of the brain,

大脑的高级中枢

例如运动皮层、小脑和基地神经节

like the motor cortex, the cerebellum, the basal ganglia,

will all modulate activity of the spinal cord

进行下行调节并产生上述信号

while we do locomotion.

并对脊髓中的活动起到调控作用

But what's interesting is to what extent just a low-level component,

但是有趣的是，脊髓和肢体这类低级中枢

the spinal cord, together with the body,

在完成运动过程中

already solve a big part of the locomotion problem.

到底扮演了何种角色

大家应该都知道，即使鸡的头被砍掉了

You probably know it by the fact that you can cut the head off a chicken,

它也不会立即死亡

it can still run for a while,

这个事实表明低级的脊髓和躯干

showing that just the lower part, spinal cord and body,

在完成运动过程中扮演了重要的角色

already solve a big part of locomotion.

Now, understanding how this works is very complex,

但是想要弄清楚它的工作原理还是很困难

because first of all,

因为，第一点就是

recording activity in the spinal cord is very difficult.

想要记录脊髓中的活动十分复杂

It's much easier to implant electrodes in the motor cortex

由于脊椎的保护作用，把电极植入到脊髓中

than in the spinal cord, because it's protected by the vertebrae.

要比植入到运动皮层中难得多

Especially in humans, very hard to do.

对于人类来说更是难上加难

A second difficulty is that locomotion is really due to a very complex

另外一个难题就是，运动基于这四个要素的相互作用

and very dynamic interaction between these four components.

是一个复杂的动态过程

So it's very hard to find out what's the role of each over time.

因此，很难区分各要素所发挥的作用

This is where biorobots like Pleurobot and mathematical models

所以，像蝾螈机器人这样的仿生机器人和数学模型

can really help.

就派上用场了

So what's biorobotics?

那么到底什么是仿生机器人学呢

Biorobotics is a very active field of research in robotics

这是机器人研究中一个热门的领域

where people want to take inspiration from animals

人们试图从动物身上获取灵感

并运用到机器人身上，使它们可以运用于户外

to make robots to go outdoors,

like service robots or search and rescue robots

例如服务用机器人、搜救机器人

or field robots.

以及农业机器人

And the big goal here is to take inspiration from animals

现在最大的目标就是从动物身上获得灵感

to make robots that can handle complex terrain --

使机器人能够应对一些复杂的地形

stairs, mountains, forests,

例如楼梯、山区、树林

places where robots still have difficulties

这些地形对于机器人来说还是个挑战

and where animals can do a much better job.

它们并不能像动物那样灵活应对

The robot can be a wonderful scientific tool as well.

机器人也是一项极好的科学研究工具

There are some very nice projects where robots are used,

在一些优秀研究项目中得以应用

like a scientific tool for neuroscience, for biomechanics or for hydrodynamics.

例如神经科学、仿生学和流体动力学

And this is exactly the purpose of Pleurobot.

这也正是蝾螈机器人存在的意义

So what we do in my lab is to collaborate with neurobiologists

所以，我们在实验室里与神经生物学家合作

like Jean-Marie Cabelguen, a neurobiologist in Bordeaux in France,

法国波尔多大学的神经生物学家 让·玛丽 卡贝肯就是其中之一

and we want to make spinal cord models and validate them on robots.

我们制作出脊髓的模型并用机器人加以验证

And here we want to start simple.

我们希望先从简单的动物入手

So it's good to start with simple animals

例如七鳃鳗

like lampreys, which are very primitive fish,

这种原始的鱼类

and then gradually go toward more complex locomotion,

然后慢慢过渡到运动更复杂的动物

like in salamanders,

例如蝾螈

but also in cats and in humans,

然后还有猫、人类

in mammals.

等哺乳动物

And here, a robot becomes an interesting tool

这样一来

机器人成为一种验证模型的有趣工具

to validate our models.

And in fact, for me, Pleurobot is a kind of dream becoming true.

实际上，蝾螈机器人的出现使得我的梦想成真

Like, more or less 20 years ago I was already working on a computer

因为，大约在20年前

我在读博士期间就通过计算机

making simulations of lamprey and salamander locomotion

模拟七鳃鳗和蝾螈的运动

during my PhD.

But I always knew that my simulations were just approximations.

但是，我一直很清楚我的模拟非常粗糙

Like, simulating the physics in water or with mud or with complex ground,

因为在水中和在泥泞等地形复杂的陆地上的运动

it's very hard to simulate that properly on a computer.

很难在电脑上精确地模拟出来

Why not have a real robot and real physics?

那么为什么不用机器人真正地模拟这些动作呢

So among all these animals, one of my favorites is the salamander.

所以在众多的动物中，蝾螈成为我最喜欢的动物之一

You might ask why, and it's because as an amphibian,

你可能会问为什么,因为从进化的角度来看

it's a really key animal from an evolutionary point of view.

蝾螈作为两栖动物,发挥着至关重要的作用

It makes a wonderful link between swimming,

它在两种运动方式间建立起美妙的联系

as you find it in eels or fish,

一个是鳗和鱼类在水中的游泳方式

and quadruped locomotion, as you see in mammals, in cats and humans.

一个是猫、人类等哺乳动物在陆上的运动方式

And in fact, the modern salamander

事实上，现代的蝾螈

is very close to the first terrestrial vertebrate,

和最早出现的陆生脊椎动物十分相近

so it's almost a living fossil,

所以说，蝾螈就像是活化石一样

which gives us access to our ancestor,

通过它，我们有机会了解我们的祖先

the ancestor to all terrestrial tetrapods.

即四足动物的祖先

So the salamander swims

蝾螈以所谓的鳗形游泳的步态

by doing what's called an anguilliform swimming gait,

在水中游动

因此肌肉活动产生的优美行波从头部传到尾部

so they propagate a nice traveling wave of muscle activity from head to tail.

And if you place the salamander on the ground,

如果把它放在地上

它的动作转变为快速爬行的步态

it switches to what's called a walking trot gait.

在这种情况下，蝾螈的四肢被定期地激活

In this case, you have nice periodic activation of the limbs

which are very nicely coordinated

十分协调地

with this standing wave undulation of the body,

随着身体驻波的起伏而运动

and that's exactly the gait that you are seeing here on Pleurobot.

这就是蝾螈机器人行走的步态的原型

Now, one thing which is very surprising and fascinating in fact

现在令人惊叹的是

is the fact that all this can be generated just by the spinal cord and the body.

所有这些活动仅靠脊髓和肢体就可以完成

如果拿一只去脑的蝾螈

So if you take a decerebrated salamander --

虽然这听起来有些残忍，但是你把它的头部切除

it's not so nice but you remove the head --

如果以低强度的刺激电激它的脊髓

and if you electrically stimulate the spinal cord,

at low level of stimulation this will induce a walking-like gait.

它会表现出类似爬行的步态

但如果你加大刺激的强度，步态就会发生变化

If you stimulate a bit more, the gait accelerates.

And at some point, there's a threshold,

这里存在一个临界值，如果强度达到临界点

and automatically, the animal switches to swimming.

它会自己变成游泳的步态

This is amazing.

这简直太神奇了

Just changing the global drive,

仅仅改变了刺激的强度

as if you are pressing the gas pedal

就像是踩下了

of descending modulation to your spinal cord,

下行调节脊髓的油门一样

实现了两种完全不同的步态之间的转变

makes a complete switch between two very different gaits.

And in fact, the same has been observed in cats.

而且我们在猫身上发现了同样的规律

If you stimulate the spinal cord of a cat,

刺激猫的脊髓

就可以使它在行走、慢跑和快跑之间进行转换

you can switch between walk, trot and gallop.

Or in birds, you can make a bird switch between walking,

还可以实现鸟从行走到拍动翅膀之间的转换

at a low level of stimulation,

相应地，需要从低强度的刺激

and flapping its wings at high-level stimulation.

转变为高强度的刺激

And this really shows that the spinal cord

这恰好证明了

脊髓是一处智能的运动控制中心

is a very sophisticated locomotion controller.

So we studied salamander locomotion in more detail,

因此我们对于蝾螈的运动展开更仔细的研究

and we had in fact access to a very nice X-ray video machine

我们从德国耶拿大学的马丁·费希尔教授的手中

from Professor Martin Fischer in Jena University in Germany.

得到了一台非常精妙的X射线成像仪

多亏了这台精妙的仪器

And thanks to that, you really have an amazing machine

我们可以记录更多关于骨骼运动的细节

to record all the bone motion in great detail.

That's what we did.

这就是我们得到的成像

So we basically figured out which bones are important for us

然后我们大致选出关键的骨骼

and collected their motion in 3D.

并且搜集他们三维运动数据

And what we did is collect a whole database of motions,

我们所做的就是搜集运动过程的全数据

both on ground and in water,

既有在陆地爬行的，也有在水中游行的

to really collect a whole database of motor behaviors

真的是动物

在运动过程中的全部数据

that a real animal can do.

And then our job as roboticists was to replicate that in our robot.

然后我们这些机器人学家的任务就是 把数据复制到机器人身上

因此我们为了找到正确的结构，进行了全套的优化方案

So we did a whole optimization process to find out the right structure,

where to place the motors, how to connect them together,

例如在哪里放置这些运动，如何让它们连贯起来

to be able to replay these motions as well as possible.

才能尽可能地重现这些动作

And this is how Pleurobot came to life.

这就是蝾螈机器人诞生的过程

So let's look at how close it is to the real animal.

接下来我们就来欣赏一下 它和活生生的动物多么相似吧

So what you see here is almost a direct comparison

你现在看到的就是非常直观的对比

between the walking of the real animal and the Pleurobot.

一个是活生生的动物，一个是蝾螈机器人

You can see that we have almost a one-to-one exact replay

你可以看到我们几乎实现了两者在步态上

of the walking gait.

一对一的重现

If you go backwards and slowly, you see it even better.

如果进行慢速回放，你会看得更清楚

But even better, we can do swimming.

但是更神奇的是，我们成功重现了游行

So for that we have a dry suit that we put all over the robot --

但我们必须给机器人穿上外套

(Laughter)

（笑声）

and then we can go in water and start replaying the swimming gaits.

然后就可以入水重现游行的步态

And here, we were very happy, because this is difficult to do.

看到这里我们很是欣慰，因为这个过程真得很艰辛

The physics of interaction are complex.

物理交互作用十分复杂

Our robot is much bigger than a small animal,

因为蝾螈机器人比真正的蝾螈大得多

so we had to do what's called dynamic scaling of the frequencies

所以我们必须进行所谓的动态频率缩放

to make sure we had the same interaction physics.

确保蝾螈机器人能实现一样的物理协调

But you see at the end, we have a very close match,

但是你看，最后两者已经十分相近

and we were very, very happy with this.

对此我们非常高兴

So let's go to the spinal cord.

让我们再回到脊髓的研究

So here what we did with Jean-Marie Cabelguen

我们和Jean-Marie Cabelguen教授所做的工作就是

is model the spinal cord circuits.

模拟脊髓中的电路

And what's interesting is that the salamander

但是非常有趣的是，我们发现蝾螈

has kept a very primitive circuit,

始终保持一种非常原始的电路

which is very similar to the one we find in the lamprey,

这和我们在七鳃鳗身上发现的很相似

this primitive eel-like fish,

就是那个原始的鳗形鱼类

and it looks like during evolution,

似乎在进化过程中

new neural oscillators have been added to control the limbs,

产生了一种新的神经振荡器来控制肢干

to do the leg locomotion.

并完成腿部的运动

And we know where these neural oscillators are

现在我们知道了这些神经振荡器的位置

but what we did was to make a mathematical model

但是我们还需要建立数学模型

to see how they should be coupled

来研究它们是如何连接起来

to allow this transition between the two very different gaits.

实现两种完全不同的步态的转换

And we tested that on board of a robot.

我们在机器人的背部进行测试

And this is how it looks.

就像这个样子

So what you see here is a previous version of Pleurobot

你现在看到的就是蝾螈机器人的原始版本

that's completely controlled by our spinal cord model

它完全由背部的脊髓模型

programmed on board of the robot.

进行控制

And the only thing we do

然后我们只需要

is send to the robot through a remote control

通过远程控制，向机器人发送

the two descending signals it normally should receive

它在正常情况下从大脑高级部位接受的到的

from the upper part of the brain.

两种下行调节信号

And what's interesting is, by playing with these signals,

有趣的是，通过这些信号

我们可以完全控制步态的速度，方向和类型

we can completely control speed, heading and type of gait.

For instance,

举例来说

when we stimulate at a low level, we have the walking gait,

当我们的刺激强度较低时，表现为行走的步态

and at some point, if we stimulate a lot,

当我们加大刺激的强度到某个极限

very rapidly it switches to the swimming gait.

它迅速转换成游行的步态

And finally, we can also do turning very nicely

最后，我们也可以很轻松地再变回来

by just stimulating more one side of the spinal cord than the other.

只需要刺激脊髓的另一端

And I think it's really beautiful

我觉得这真的很美妙

how nature has distributed control

大自然竟然赋予了脊髓

to really give a lot of responsibility to the spinal cord

这么重大的责任

以至于脑部高级中枢部分根本不需要担心每一块肌肉

so that the upper part of the brain doesn't need to worry about every muscle.

It just has to worry about this high-level modulation,

它只需要进行高层调节

and it's really the job of the spinal cord to coordinate all the muscles.

而协调所有的肌肉就是脊髓的任务了

So now let's go to cat locomotion and the importance of biomechanics.

我们再回到猫的运动，进一步认识仿生学的的重要性

So this is another project

这就是另一个项目了

where we studied cat biomechanics,

我们对猫进行了仿生学研究

and we wanted to see how much the morphology helps locomotion.

想弄清楚形态学对于肢体动作的影响

And we found three important criteria in the properties,

我们发现，从根本上讲

躯干具有三个重要的属性

basically, of the limbs.

The first one is that a cat limb

首先就是猫的四肢，

more or less looks like a pantograph-like structure.

或多或少地类似于受电弓的结构

So a pantograph is a mechanical structure

受电弓是一种机械结构

which keeps the upper segment and the lower segments always parallel.

它使得上框架和下框架始终保持平行

So a simple geometrical system that kind of coordinates a bit

如此简单的几何学系统

the internal movement of the segments.

竟能协调上下框架的内部移动

A second property of cat limbs is that they are very lightweight.

猫的四肢的第二个属性就是轻盈性

Most of the muscles are in the trunk,

大部分的肌肉都集中在身体

which is a good idea, because then the limbs have low inertia

这真是个好点子，这样一来四肢的惯性就很小

and can be moved very rapidly.

猫就可以灵活地运动

The last final important property is this very elastic behavior of the cat limb,

最后一个重要的属性是四肢的弹性

so to handle impacts and forces.

可以应对各种冲击和压力

And this is how we designed Cheetah-Cub.

我们就是根据这些属性设计出猎豹机器人

So let's invite Cheetah-Cub onstage.

接下来有请猎豹机器人登场

So this is Peter Eckert, who does his PhD on this robot,

这是由彼得埃克特在博士期间研制的机器人

and as you see, it's a cute little robot.

你可以看到，它是一个非常可爱的小型机器人

It looks a bit like a toy,

看起来就像玩具一样

but it was really used as a scientific tool

但它的的确确是一项科研工具

to investigate these properties of the legs of the cat.

用来研究猫的四肢的属性

So you see, it's very compliant, very lightweight,

你看，它的四肢十分协调、轻盈

and also very elastic,

并且富有弹性

so you can easily press it down and it will not break.

你可以轻而易举地把它压下去，但它一点事都没有

It will just jump, in fact.

实际上，它只会跳起来

And this very elastic property is also very important.

所以说四肢的弹跳性非常重要

And you also see a bit these properties

你也可以把四肢的三个节段

of these three segments of the leg as pantograph.

看作一个受电弓

Now, what's interesting is that this quite dynamic gait

现在，有趣的是这个动态的装置

is obtained purely in open loop,

实现了完全的开环回路

meaning no sensors, no complex feedback loops.

也就是说没有传感器，也没有复杂的反馈回路

And that's interesting, because it means

这真的很有趣，因为这意味着

that just the mechanics already stabilized this quite rapid gait,

仅仅靠机器就可以维持这个敏捷的装置的平衡

and that really good mechanics already basically simplify locomotion.

并且这个精巧的机器对运动进行了基本的简化

To the extent that we can even disturb a bit locomotion,

在某种程度上，我们甚至可以在运动过程中制造一些障碍

as you will see in the next video,

在接下来的视频是你会看到

where we can for instance do some exercise where we have the robot go down a step,

例如我们做一些实验让机器人下台阶

and the robot will not fall over,

机器人并不会摔倒

which was a surprise for us.

这确实出乎我们的意料

This is a small perturbation.

对于这个细微的干扰

I was expecting the robot to immediately fall over,

我预期机器人会立刻摔倒

because there are no sensors, no fast feedback loop.

因为它没有传感器，也没有快速反馈回路

But no, just the mechanics stabilized the gait,

但是出乎我们的意料，仅仅靠这个机器维持了装着的平衡

and the robot doesn't fall over.

机器人并没有摔倒

Obviously, if you make the step bigger, and if you have obstacles,

显然，如果增大跨度，再设置一些障碍

you need the full control loops and reflexes and everything.

就需要全控制回路和反射弧等一系列东西

But what's important here is that just for small perturbation,

但是眼下最重要的是，对于一些细微的干扰

the mechanics are right.

仅仅靠机器就可以完成

And I think this is a very important message

我想这对仿生学家、机器人学家和神经学家来说

from biomechanics and robotics to neuroscience,

是一个非常重要的信息，那就是：

saying don't underestimate to what extent the body already helps locomotion.

不要低估躯干对于运动所发挥的作用

Now, how does this relate to human locomotion?

那么这是如何联系到人类的运动上的呢

Clearly, human locomotion is more complex than cat and salamander locomotion,

显然，人类的运动比猫和蝾螈的运动更为复杂

but at the same time, the nervous system of humans is very similar

但是另一方面，人类的神经系统

和其他脊椎动物的十分相似

to that of other vertebrates.

And especially the spinal cord

尤其是脊髓

is also the key controller for locomotion in humans.

它同样是人类运动的重要处理中心

That's why, if there's a lesion of the spinal cord,

这就是为什么一旦脊髓受损

this has dramatic effects.

将会导致严重的后果

The person can become paraplegic or tetraplegic.

例如瘫痪或者四肢麻痹

This is because the brain loses this communication

这是由于大脑与脊髓

with the spinal cord.

之间的信息传递受损

Especially, it loses this descending modulation

尤其是用于引发和调节运动的

to initiate and modulate locomotion.

下行调节机制

So a big goal of neuroprosthetics

所以神经学家的一大目标

is to be able to reactivate that communication

就是通过电刺激或者化学刺激

using electrical or chemical stimulations.

激活两者之间的信息传递

And there are several teams in the world that do exactly that,

世界上有些团队就在研究这一领域

especially at EPFL.

在洛桑联邦理工学院更是如此

My colleagues Grégoire Courtine and Silvestro Micera,

例如我的同事格雷瓜尔·库尔蒂纳

with whom I collaborate.

还有西尔维斯特·米切拉

But to do this properly, it's very important to understand

但是为了做好这一点，理解以下几点非常重要

how the spinal cord works,

脊髓是如何工作的

how it interacts with the body,

它是如何与躯干相互作用的

and how the brain communicates with the spinal cord.

以及大脑是如何与脊髓进行信息传递的

This is where the robots and models that I've presented today

我今天所呈现的机器人和模型

will hopefully play a key role

就将针对上述重要问题

towards these very important goals.

发挥关键的作用

Thank you.

感谢大家

(Applause)

（掌声）

Bruno Giussani: Auke, I've seen in your lab other robots

布鲁诺·朱萨尼：奥克，我在你的实验室 看到一些其他的机器人

that do things like swim in pollution

例如在污水中游泳

and measure the pollution while they swim.

并检测污染程度的机器人

But for this one,

但是你在演讲中提到的蝾螈机器人

you mentioned in your talk, like a side project,

它更像是一个小项目

search and rescue,

它的鼻子上有一个照相机

and it does have a camera on its nose.

可以进行搜救工作

Auke Ijspeert: Absolutely. So the robot --

奥克：确实是这样，关于机器人

We have some spin-off projects

我们有很多派生项目

where we would like to use the robots to do search and rescue inspection,

我们希望用这些机器人进行搜救和视察工作

so this robot is now seeing you.

所以蝾螈机器人诞生了

And the big dream is to, if you have a difficult situation

我们设想的是，如果有人陷入困境

like a collapsed building or a building that is flooded,

例如建筑坍塌或者遭遇洪水

and this is very dangerous for a rescue team or even rescue dogs,

这对救援团队和搜救犬来说都是危险的

why not send in a robot that can crawl around, swim, walk,

那么为什么不派一个个既会爬行 又会游行的机器人呢

with a camera onboard to do inspection and identify survivors

它的背部还装有照相机，可以用来 进行侦查、识别幸存者

and possibly create a communication link with the survivor.

甚至可能实现与幸存者的联系

BG: Of course, assuming the survivors don't get scared by the shape of this.

布鲁诺:当然，前提是幸存者没有被机器人的样子吓到

AI: Yeah, we should probably change the appearance quite a bit,

奥克:是的，我们将对机器人的外形进行改善

because here I guess a survivor might die of a heart attack

否则，我猜幸存者可能会被吓出心脏病

just of being worried that this would feed on you.

害怕机器人会吃了他们

But by changing the appearance and it making it more robust,

但是我相信，通过改变它的外形，让它更强健一些

I'm sure we can make a good tool out of it.

它将会发挥很重要的作用

布鲁诺:非常感谢你和你的团队

BG: Thank you very much. Thank you and your team.