-
Not Synced
This is Pleurobot.
-
Not Synced
Pleurobot is a robot that we designed
to closely mimic a salamander species
-
Not Synced
called ??
-
Not Synced
Pleurobot can walk, as you can see here,
-
Not Synced
and as you'll see later, it can also swim.
-
Not Synced
So you might ask,
why did we design this robot?
-
Not Synced
And in fact, this robot has been designed
as a scientific tool for neuroscience.
-
Not Synced
Indeed, we designed it
together with neurobiologists
-
Not Synced
to understand how animals move,
-
Not Synced
and especially how the spinal cord
controls locomotion.
-
Not Synced
But the more I work in biorobotics,
-
Not Synced
the more I'm really impressed
by animal locomotion.
-
Not Synced
If you think of a dolphin swimming
or a cat running or jumping around,
-
Not Synced
or even us as humans,
-
Not Synced
when you go jogging or play tennis,
-
Not Synced
we do amazing things.
-
Not Synced
And in fact, our nervous system solves
a very, very complex control problem.
-
Not Synced
It has to coordinate more
or less 200 muscles perfectly,
-
Not Synced
because if the coordination is bad,
we fall over or we do bad locomotion.
-
Not Synced
And my goal is to understand
how this works.
-
Not Synced
There are four main components
behind animal locomotion.
-
Not Synced
The first component is just the body,
-
Not Synced
and in fact we should never underestimate
what extent the biomechanics
-
Not Synced
already simplify locomotion in animals.
-
Not Synced
Then you have the spinal cord,
-
Not Synced
and in the spinal cord you find reflexes,
-
Not Synced
like multiple reflexes that create
a sensory motor coordination loop
-
Not Synced
between neural activity in the spinal cord
and mechanical activity.
-
Not Synced
A third component
are central pattern generators.
-
Not Synced
These are very interesting circuits
in the spinal cord of vertebrate animals
-
Not Synced
that can generate, by themselves, very
coordinated rhythmic patterns of activity
-
Not Synced
while receiving only
very simple input signals.
-
Not Synced
And these input signals come from
descending modulation
-
Not Synced
from higher parts of the brain,
from the motor cortex, the cerebellum,
-
Not Synced
the basal ganglia, will all modulate
activity of the spinal cord
-
Not Synced
while we do locomotion.
-
Not Synced
But what's interesting is to what extent
just a low level component,
-
Not Synced
the spinal cord, together with the body,
already solves a big part
-
Not Synced
of the locomotion problem,
-
Not Synced
and you probably know it by the fact
that you can cut the head of the chicken,
-
Not Synced
it can still run for a while,
-
Not Synced
showing that just the lower part,
spinal cord and body,
-
Not Synced
already solves a big part of locomotion.
-
Not Synced
Now, understanding how this works
is very complex,
-
Not Synced
because first of all,
-
Not Synced
recording activity in the spinal cord
is very difficult.
-
Not Synced
It's much easier to implant electrodes
in the motor cortex
-
Not Synced
than in the spinal cord, because
it's protected by the vertebrae.
-
Not Synced
Especially in humans,
it's very hard to do.
-
Not Synced
A second difficulty is that locomotion
is really due to a very complex
-
Not Synced
and very dynamic interaction
between these four components.
-
Not Synced
So it's very hard to find out
what's the role of each over time.
-
Not Synced
This is where biorobots like Pleurobot
and mathematical models
-
Not Synced
can really help.
-
Not Synced
So what's biorobotics?
-
Not Synced
Biorobotics is a very active field
of research in robotics
-
Not Synced
where people want to take inspiration
from animals to make robots
-
Not Synced
to go outdoors,
-
Not Synced
like service robots
or search-and-rescue robots
-
Not Synced
or field robots,
-
Not Synced
and the big goal here is
to take inspiration from animals
-
Not Synced
to make robotics that can handle
complex terrain --
-
Not Synced
stairs, mountains, forests,
-
Not Synced
places where robots
still have difficulties
-
Not Synced
and where animals can do
a much better job.
-
Not Synced
The robot can be
a wonderful scientific tool as well.
-
Not Synced
There are some very nice projects
where robots are used
-
Not Synced
like a scientific tool for neuroscience,
for biomechanics, or for ?? dynamics.
-
Not Synced
And this is exactly
the purpose of Pleurobot.
-
Not Synced
So what we do in my lab
is to collaborate with neurobiologists
-
Not Synced
like Jean-Marie Cabelguen,
a neurobiologist in Bordeaux in France,
-
Not Synced
and we want to make spinal cord models
and validate them on robots.
-
Not Synced
And here we want to start simple.
-
Not Synced
So it's good to start with simple animals
-
Not Synced
like lampreys, which are
very primitive fish,
-
Not Synced
and then gradually go toward
more complex locomotion,
-
Not Synced
like in salamanders,
-
Not Synced
but also in cats and in humans,
-
Not Synced
in mammals.
-
Not Synced
And here, a robot becomes
an interesting tool
-
Not Synced
to validate our models,
-
Not Synced
and in fact, for me, Pleurobot
is a kind of dream becoming true.
-
Not Synced
Like, more or less 20 years ago
I was already working on a computer
-
Not Synced
making simulations of lamprey
and salamander locomotion
-
Not Synced
during my Ph.D.
-
Not Synced
But I always knew that my simulations
were just approximations.
-
Not Synced
Like, simulating the physics in water
or with mud or with complex ground,
-
Not Synced
it's very hard to simulate that
properly on a computer.
-
Not Synced
Why not have a real robot
and real physics?
-
Not Synced
So among all these animals,
one of my favorites is the salamander.
-
Not Synced
You might as why, and it's because
as an amphibian,
-
Not Synced
it's a really key animal
from an evolutionary point of view.
-
Not Synced
It makes a wonderful link
-
Not Synced
between swimming, as you find it
in eels or fish,
-
Not Synced
and quadruped motion, as you see
in mammals, in cats and humans.
-
Not Synced
And in fact, the modern salamander
-
Not Synced
is very close to the first
terrestrial vertebrate,
-
Not Synced
so it's almost a living fossil,
-
Not Synced
which gives us access to our ancestor,
-
Not Synced
the ancestor to all terrestrial tetrapods.
-
Not Synced
So the salamander swims by doing
what's using what's called
-
Not Synced
a ??? swimming gait,
-
Not Synced
so they propagate a nice traveling wave
of muscle activity from head to tail.
-
Not Synced
And if you place the salamander
on the ground,
-
Not Synced
it switches to what's called
a walking trot gait.
-
Not Synced
In this case, you have nice
activation of the limbs
-
Not Synced
which are very nicely coordinated
with this standing wave undulation
-
Not Synced
of the body,
-
Not Synced
and that's exactly the gait
that you are seeing here on Pleurobot.
-
Not Synced
Now, one thing which is very surprising
and fascinating in fact
-
Not Synced
is the fact that all this can be generated
just by the spinal cord and the body.
-
Not Synced
So if you take ?? salamander --
-
Not Synced
it's not so nice
but you remove the head --
-
Not Synced
and if you electrically stimulate
the spinal cord,
-
Not Synced
a low level of stimulation
this will use a walking-like gait.
-
Not Synced
If you stimulate a bit more,
the gait accelerates,
-
Not Synced
and at some point, there's a transfer,
and automatically,
-
Not Synced
the animal switches to swimming.
-
Not Synced
This is amazing,
-
Not Synced
just changing the global drive
as if you are pressing the gas pedal
-
Not Synced
of descending modulation
to your spinal cord,
-
Not Synced
makes a complete switch
between two very different gaits.
-
Not Synced
And in fact, the same
has been observed in cats.
-
Not Synced
If you stimulate the spinal cord of a cat,
-
Not Synced
you can switch between
walk, trot, and gallop.
-
Not Synced
Or in birds, you can make a bird
switch between walking,
-
Not Synced
at low levels of stimulation,
-
Not Synced
and flapping its wings
at high level stimulation.
-
Not Synced
And this really shows that the spinal cord
is a very sophisticated
-
Not Synced
locomotion controller.
-
Not Synced
So we studied salamander locomotion
in more detail,
-
Not Synced
and we had in fact access
to a very nice x-ray video machine
-
Not Synced
from Professor Martin Fischer
in Jena University in Germany.
-
Not Synced
And thanks to that, you really have
an amazing machine
-
Not Synced
to record all the bone motion
in great detail.
-
Not Synced
That's what we did.
-
Not Synced
So we basically figured out
which bones are important for us
-
Not Synced
and collected their motion in 3D.
-
Not Synced
And what we did is collect
a whole database of motions,
-
Not Synced
both on ground and in water,
-
Not Synced
to really collect the whole database
of motor behaviors
-
Not Synced
that a real animal can do,
-
Not Synced
and then our job as roboticists
was to replicate that in our robot.
-
Not Synced
So we did a whole optimization process
to find out the right structure,
-
Not Synced
where to place the motors,
how to connect it together,
-
Not Synced
to be able to replay
these motions as well as possible.
-
Not Synced
And this is how Pleurobot came to life.
-
Not Synced
So let's look how closely it is
to the real animal.
-
Not Synced
So what you see here is almost
a direct comparison
-
Not Synced
between the walking
of the real animal and the Pleurobot.
-
Not Synced
You can see that we have almost
A one-to-one exact replay
-
Not Synced
of the walking gait.
-
Not Synced
If you go backwards and slowly,
you see it even better.
-
Not Synced
But even better, we can do swimming.
-
Not Synced
So for that we have a dry suit
that we put all over the robot --
-
Not Synced
(Laughter) --
-
Not Synced
and then we can go in water
and start replaying the swimming gaits.
-
Not Synced
And here, we were very happy,
because this is difficult to do.
-
Not Synced
The physics of interaction are complex.
-
Not Synced
Our robot is much bigger
than a small animal,
-
Not Synced
so we had to do what's called
dynamical scaling of the frequencies
-
Not Synced
to make sure we had the same
interaction physics.
-
Not Synced
But you see at the end
we have a very close match,
-
Not Synced
and we were very, very happy with this.
-
Not Synced
So let's go do the spinal cord.
-
Not Synced
So here what we did
with Jean-Marie Cabelguen
-
Not Synced
is model the spinal cord circuits.
-
Not Synced
And what's interesting is that
the salamander has kept
-
Not Synced
a very primitive circuit
-
Not Synced
which is very similar to the one
we find in the lamprey,
-
Not Synced
pretty much this eel-like fish,
-
Not Synced
and it looks like during evolution,
-
Not Synced
new neuronal oscillators have been added
to control the limbs,
-
Not Synced
to do the leg locomotion.
-
Not Synced
And we know where
these neuronal oscillators are
-
Not Synced
but what we did was to make
a mathematical model
-
Not Synced
to see how they should be coupled
-
Not Synced
to allow this transition between
the two very different gaits.
-
Not Synced
And we tested that on board of a robot.
-
Not Synced
And this is how it looks.
-
Not Synced
So what you see here
-
Not Synced
is a previous version of Pleurobot
-
Not Synced
that's completely controlled
by our spinal cord model
-
Not Synced
programmed on board of the robot.
-
Not Synced
And the only thing we do
is send to the robot
-
Not Synced
through a remote control
-
Not Synced
the two descending signals
it normally should receive
-
Not Synced
from the upper part of the brain.
-
Not Synced
And what's interesting is,
by playing with these signals,
-
Not Synced
we can completely control
speed, heading, and type of gait.
-
Not Synced
For instance,
-
Not Synced
when we stimulate at a low level,
we have the walking gait,
-
Not Synced
and at some point, if we stimulate a lot,
very rapidly it switches
-
Not Synced
to the swimming gait.
-
Not Synced
And finally, we can also do turning
-
Not Synced
very nicely by just stimulating more one
side of the spinal cord than the other.
-
Not Synced
And I think it's really beautiful
-
Not Synced
how nature has distributed control
-
Not Synced
to really give a lot of responsibility
to the spinal cord
-
Not Synced
so that the upper part of the brain
doesn't need to worry about every muscle.
-
Not Synced
It just has to worry about
this high-level modulation,
-
Not Synced
and it's really the job of the spinal cord
to coordinate all the muscles.
-
Not Synced
So now let's go to cat locomotion,
and the importance of biomechanics.
-
Not Synced
So this is another project
-
Not Synced
where we studied cat biomechanics,
-
Not Synced
and we wanted to see how much
the morphology helps locomotion.
-
Not Synced
And we found three important
criteria in the properties,
-
Not Synced
basically, of the limbs.
-
Not Synced
The first one is that a cat limb
-
Not Synced
more or less looks
like a pantograph-like structure.
-
Not Synced
So a pantograph is a mechanical structure
-
Not Synced
which keeps the upper segment
and the lower segments always parallel.
-
Not Synced
So a simple geometrical system
that kind of coordinates a bit
-
Not Synced
the internal movement of the segments.
-
Not Synced
A second property of cat limbs
is that they are very lightweight.
-
Not Synced
Most of the muscles are in the trunk,
-
Not Synced
which is a good idea, because then
the limbs have low inertia
-
Not Synced
and can be moved very rapidly.
-
Not Synced
The last final important property is this
very elastic behavior of the cat limb,
-
Not Synced
so to handle impacts and forces.
-
Not Synced
And this is how we designed Cheetah-Cub.
-
Not Synced
So let's invite Cheetah-Cub onstage.
-
Not Synced
So this is Peter Eckert, who does
his Ph.D on this robot,
-
Not Synced
and as you see, it's a cute little robot.
-
Not Synced
It looks a bit like a toy,
-
Not Synced
But it was really used
as a scientific tool
-
Not Synced
to investigate these properties
of the legs of the cat.
-
Not Synced
So you see, it's very compliant,
very lightweight,
-
Not Synced
and also very elastic,
-
Not Synced
so you can easily press it down
and it will not break.
-
Not Synced
It will just jump, in fact.
-
Not Synced
And this very elastic property
is also very important.
-
Not Synced
And you also see a bit this properties
of these three segments
-
Not Synced
of the leg as pantograph.
-
Not Synced
Now, what's interesting is that
this quite dynamic gait
-
Not Synced
is obtained purely in open loop,
-
Not Synced
meaning no sensors,
no complex feedback loops.
-
Not Synced
And that's interesting, because it means
-
Not Synced
that just the mechanics already stabilized
this quite rapid gait,
-
Not Synced
and that really good mechanics
already basically simplify locomotion.
-
Not Synced
To the extent that we can even
disturb a bit locomotion,
-
Not Synced
as you will see in the next video,
-
Not Synced
where we can for instance do some exercise
where we have the robot go down a step,
-
Not Synced
and the robot will not fall over,
-
Not Synced
which was a surprise for us.
-
Not Synced
This is a small perturbation.
-
Not Synced
I was expecting the robot
to immediately fall over,
-
Not Synced
because there is no sensors,
no fast feedback loop.
-
Not Synced
But no, just the mechanics
stabilized the gait,
-
Not Synced
and the robot doesn't fall over.
-
Not Synced
Obviously, if you make the step bigger,
and if you have obstacles,
-
Not Synced
you need the full control loops
and reflexes and everything,
-
Not Synced
but what's important here
is that just for small perturbation,
-
Not Synced
the mechanics are right.
-
Not Synced
And I think this is
a very important message
-
Not Synced
from biomechanics and robotics
to neuroscience,
-
Not Synced
saying don't underestimate to what extent
-
Not Synced
the body already helps locomotion.
-
Not Synced
Now, how does this relate
to human locomotion?
-
Not Synced
Clearly, human locomotion is more complex
than cat and salamander locomotion,
-
Not Synced
but at the same time, the nervous system
of humans is very similar
-
Not Synced
to that of other vertebrates.
-
Not Synced
But at the same time, the nervous system
of humans is very similar
-
Not Synced
to that of other vertebrates,
-
Not Synced
and especially the spinal cord is also the
key controller for locomotion in humans.
-
Not Synced
That's why, if there's a lesion
of the spinal cord,
-
Not Synced
this has dramatic effects.
-
Not Synced
The person can become
paraplegic or tetraplegic.
-
Not Synced
This is become the brain
loses its communication
-
Not Synced
with the spinal cord.
-
Not Synced
Especially, it loses
the descending modulation
-
Not Synced
to initiate and modulate locomotion.
-
Not Synced
So a big goal of neuroprosthetics
-
Not Synced
is to be able to reactivate
that communication
-
Not Synced
using electrical or chemical stimulations.
-
Not Synced
And there are several teams
in the world that do exactly that,
-
Not Synced
especially at EPFL.
-
Not Synced
My colleagues ?? and ??,
-
Not Synced
with whom I collaborate.
-
Not Synced
But to do this properly,
it's very important to understand
-
Not Synced
how the spinal cord works,
-
Not Synced
how it interacts with the body,
-
Not Synced
and how the brain communicates
with the spinal cord.
-
Not Synced
This is where the robots
and models that I've presented today
-
Not Synced
will hopefully play a key role
-
Not Synced
towards these very important goals.
-
Not Synced
Thank you.
-
Not Synced
(Applause)
-
Not Synced
Bruno Giussani: Okay, I've seen
in your lab other robots
-
Not Synced
that do things like swim in pollution
and measure the pollution
-
Not Synced
while they swim,
-
Not Synced
but for this one, you mentioned
in your talk, like a side project,
-
Not Synced
search and rescue,
-
Not Synced
and it does have a camera on its nose.
-
Not Synced
Auke Ijspeert: Absolutely. So the robot,
we have some spinoff projects
-
Not Synced
where we would like to use the robots
to do search and rescue inspection,
-
Not Synced
so this robot is now seeing you.
-
Not Synced
And the big dream is to,
if you have a difficult situation
-
Not Synced
like a collapsed building
or a building that is flooded,
-
Not Synced
and this is very dangerous
for a rescue team or even rescue dogs,
-
Not Synced
why not send in a robot
that can crawl around, swim, walk,
-
Not Synced
with a camera onboard
to do inspection and identify survivors
-
Not Synced
and possibly create a communication link
with the survivor.
-
Not Synced
BG: Of course, assuming the survivors
don't get scared by the shape of this.
-
Not Synced
AI: Yeah, we should probably change
the appearance quite a bit,
-
Not Synced
because here I guess a survivor
might die of a heart attack
-
Not Synced
just by being worried that this
would feed on you.
-
Not Synced
But by changing the appearance
and it making it more robust,
-
Not Synced
I'm sure we can make
a good tool out of it.
-
Not Synced
BG: All right, thank you very much.
Thank you and your team.
-
Not Synced
(Applause)