Right.

I get this.

Listen.

Were fired.

And.

Why.

Then thanks very much and thanks to
everyone for making that such a great day.

I really had a fabulous time.

Georgia Tech very inspiring.

There's a lot of great faculty and
students here and

I'm OK I've got lots of new ideas to.

To take back to the to the lab.

Now as Dan alluded to I was trained
as an evolutionary biologist I

didn't know any years ago
a robot from anything.

But over the years as I have as
I've worked more in the form and

function evolution of fishes.

I thought in order to
try to understand and

answer some of the questions
I want to answer.

I need to take a robotic approach.

So I'll start the talk today with a little
bit of a background on fishes and

fish evolution fish diversity but
not too much.

I'll tell you a little bit about some
of the questions we're interested in

answering and

then some of the different robotic
systems that were developed ourselves or

in conjunction with collaborators who
are using to answer those questions and

because I know many of you don't
work on the aquatic realm with fish.

I'm going to keep this fairly general and
just I'll show so

relatively more movies relatively more
pictures relatively less data but

certainly you're welcome
to a look at the web page.

We have all of our publications up and
you can probe me via e-mail.

If you would like to ask more questions
about any of the specific areas.

What so to begin with.

I'll just say that someone is
instead invertebrate biology.

This is a are great.

Model system there represent half of the
diversity of fishes there are somewhere on

the order of sixty thousand vertebrates
that is birds mammals and Fabians reptiles

fish fish now is slightly over thirty
thousand the total number of species so

the the diversification that's taken
place over more than half a billion years

in the aquatic realm is quite impressive
fish of really radiated into the three

dimensional habitat represented by the
water in almost every imaginable way and

they've had many years to have selection
operate on their propulsive systems

to involve multiple propulsive multiple
surfaces that push on the water

devolve flexible proposers to evolve
highly maneuverable body plans that enable

them to maneuver in and out of Coral Reef
heads obstacles in the water to deal with

turbulent flows and so forth and there's
so an obvious really a source of its

inspiration for understanding
the aquatic around how we might

build human devices that could move
around the water and locomotive.

In a way that might eventually
approximate what what fish can do.

But I think we're still really at
the very beginning of this process so

I emphasize especially since
it Georgia Tech and there's so

many people going to
rest year old robots and

some flight work that we're way behind
I think on in the aquatic arena.

What's been accomplished on land on sand.

And for

terrestrial locomotion in general but
we're making progress but if we have it.

We have a long way to go.

What I'd like to do today is to begin with
a few minutes about fish some of the major

evolutionary trends some of
the major design features that fish

exhibit as part of their
locomotor a function.

And then do a sort of a survey
if you like of different fish

robotic platforms that
we've worked on in my lab.

I'll talk about some of
the different approaches.

We've taken present briefly some of the
more complex biomimetic robotic devices

spend the bulk of the girl.

A significant portion of time on these
very flexible simple bending models for

a swimming fish.

And then wrap up with some of
our most recent work using

these robotic devices to investigate some
interesting features of the structure and

function of fish which in this case
would be the structure of sharkskin So

some of you may know sharkskin has
a very particular structure it served as

inspiration for a number of
different engineering materials and

surfaces and I'll tell you a little bit
about about how we've studied that.

And I can't continue without really giving
tremendous praised all my collaborators

at various universities.

James Tate Gora directs all
sort of chief among them but

Silas all been here at Georgia Tech's been
an important computational collaborator.

Eric Anderson Brock Malcolm at
Northwestern University harness and

we're been for Mendus in addition
to a large number of students so

being trained as a as an evolutionary
biologist I never could have done

this work without some
great collaborators.

Now one of the key obvious features of the
Georgia Aquarium right nearby many of you

I'm sure have been there of the diversity
if this is the diversity of body shapes

tale shapes the shapes that we see fish
that have a long date fins here that

they used to swim to maneuver fish with
a variety of different fins arrayed

around the body fish with symmetrical tail
space with asymmetrical tails we've worked

quite a bit on that although I won't
talk about that today to know trout and

salmon and so forth but
to understand the diversity of this.

I think we can compare a tuna to a shark.

But there's only so much.

I think we could learn we can
learn a lot of biology but

if we want to isolate
an individual trait and

say well what's the this tail shape mean
in terms of local motor performance.

It's very hard to approach that from
a comparative point of view alone.

I think we need some way to
control this diversity and

manipulate it in the lab and
a robotic approach really allow.

Us to do that.

So we've tried to study it comparatively
in fact these are all species that we've

worked on or tuna relatives anyway but
also with robotic models and

one one form or another.

Now that trends in the evolution of
fishes have been well documented for

many years and
not just review some of them.

For you in this very simplified
evolutionary diagram of fishes so

there are thirty thousand species of fish
that are represented on this diagram.

It's obviously a little bit simplified
the RAIF in fishes is the group that.

I'm going to show you here.

So we're excluding the sharks and
rays which would come off at the base but

when we look at the evolution of pectoral
fans and pelvic fins which I'm showing

here in purple and yellow we can
see these early evolutionary groups

like sturgeon have a very low on the body
wing like a pectoral fan as you kind

of move up the phylogeny you wind up
with spectral fins located right under

the center of mass which is about right
here up on the side of the body and

they're much more flexible and new verbal
in three dimensions that is this fish here

can move this Petrel thinned quite a bit
and in three dimensions sturgeon is much

more limited in what it can do we see many
trends in pelvic thin so here the pelvic

bones are in yellow they tend to move
forward again winding up under the center

of mass or nearly under the center of mass
and more derived evolutionary groups.

And we see trends in the dorsal and
anal fins shown here in red and blue.

There are a number of patterns
these fins can be right on top

of one another they can be offset by this.

There's a major group of fishes called
the spiny fin fish is where the spines.

The dorsal fin here in front
is a spiny in nature and

sometimes they're set separated with a
soft region flexible and a spiny part and

sometimes it's attached
like we see here and

here we see another case where the two
dorsal and the fans are right right.

Vertically arranged with one another.

And finally we have trends in
the evolution of the tail.

Where we have.

Early in evolution this asymmetrical shark
like tail with an upper lobe that's bigger

and extends further out than the lower
lobe at the origin of the true bony

fishes you tend to have a symmetrical
tail and have many shapes.

But the upper lobe is very similar to
the lower lobe in area motions of those

lobes could be different but they're
nonetheless externally symmetrical we can

have four tails and tails where there's
a relatively straight trailing edge.

So all of these patterns are.

Interesting.

They've been noted by paleontologists for
years but

we have almost no idea what they mean.

So what does it mean to have fins in
different positions what does it mean to

have a tail like this versus a tail like
that if we want to have a stiffer fish

versus a more flexible fish What does that
mean terms of locomotor function if we

have a fish it's longer versus the fish
the shorter That's a very simple

morphological change what does that mean
what about the hydrogen and function

of these different types of fins where we
really have very little idea about that.

And so over the past years in my lab.

Last ten years or so
we've been working on fish to try and

understand what it means
to have a tail like that

image the flow image of the motion and now
what I'll be telling you about today is

to build some robotic models to assess
different aspects of fish bin function.

Now just a quick overview.

Of fins here in fishes so

here is my favorite model system actually
the bluegill sunfish there are a whole

series of fins there's the dorsal fin
This is the spiny part this is flexible.

There's the tail.

There's the anal fin of
these are mid-line fins.

And then there are the paired fins so
they're at the pectoral fins here one

on this side one on the other and
the pair pelvic fins down here.

And these are all under active control
by the fish they're then in muscles that

control these these fans may have
an interesting design that we'll talk

about in a second in terms of the whole

fish one of the interesting things about
fish is that most of them are unstable.

That is their center of buoyancy
is below the center of mass.

So if you anesthetize this fish
it would turn upside down and

SAP is slightly negatively buoyant so
it would sink to the bottom.

So as a control problem you can see right
away that we've ignored in the fish

commotion community for the most part
this fish needs to actively move its fins

to remain stable fish can certainly rotate
along their long absence Here's a perch

turning surface zero radius turn if you'd
like to using its pectoral fins and

a little bit of the tail back here.

But this need for control of body
posture is certainly been a driving

force in fish evolution is something
I won't really discuss today but

it's a very interesting area for future
work and the last thing I'll say about

the functional design officious is how
the fins themselves are constructed so

some of you have seen this side
before I'll show it again for

those of you haven't and
present a very simple model of the fin.

Here's the fin of the fish.

The fin is composed of these individual
rays that stick out from the base of

the fin that are actually made of jointed
small bones and there's a thin membrane.

In between those bones we've worked on the
mechanical properties of this a little bit

to try and better understand what
the what the how the system works.

Mechanically.

But one feature the most general
interesting feature I think for

today is that each one of these rays here
that supports the fit is composed of two

parts to there are two half rays and
their muscles at the base of

this to control them in fact in each fin
Ray here there are four muscles so if

you have a fin Ray that's got a fin it's
got fourteen rays you have on the order of

sixty muscles that control that and
we've done some muscle recordings of these

to show how the fin is working but we're
nowhere near getting sixty simultaneous

recordings from the muscles so it's
a very complex control problem here too.

The interesting thing from the mechanical
point of view is if you asymmetrically

displace the bases of these fin rays
what you see is that the thinner.

Curves and so in contrast to insect
flight or bird feathers for example fish

have the ability to actively control the
curvature of their appendage by muscles at

the base and my simple model for that
some of you have seen is a Ziploc bag.

So I've cut open the top
of the Ziploc bag.

I've opened up the base.

And if you zip it together and

make your fingers do the work
of the muscles at the base.

You can see the curvature of that
the fish can apply and from a hundred and

human point of view it's very interesting
because oncoming flows normally would bend

the fin back against the flow but fish
can resist this and then their appendage

into the flow actively resisting
the defamation caused by the fluid.

It makes determining what's active and

passive defamation assuming this very
difficult but it's a very interesting

mechanical system I'm not aware of many
examples like that in the animal kingdom.

I won't actually say much more about
it because we've built some early

robots with this device and
it doubles the number of actuators and so

we've temporarily abandoned that approach.

Now let's turn to the diversity of
robotic platforms that we have this

diversity of fishes at different
shapes different things.

Structures we don't really have that
much of a sense of what it means how

might we attack that from
a robotic point of view.

So what we've tried to do over the last
five or six years so now is to try

to apply a diversity of approaches
something very simple to something much

more complex and biomimetic that we're
approaching now show you some examples.

Not this one in particular but another one
in a moment the biomimetic designs have

some nice features they
accurately mimic the biology.

You can program natural fin motions
It looks sort of like a fish.

You can make alterations in
the structure of these fin rays you

can measure the forces but the downside is
it takes a long time to build one of these

that works well and we're still working
on this particular one right here.

Also some slides in a moment.

So they're hard to build and control.

In the middle.

We.
Individual fin elements that I will

show you some information from that are
meant to serve as a stepping stone to this

robot here so
we'll talk about pectoral fins and

tail fin robots and
then on the other extreme.

We have some very simple physical
models of swimming fish that I didn't

think at first were going to
be all that informative but

it turned out to be extremely interesting
and these are just simple flexing bending

pieces of plastic that
will discuss later on.

That's going to these
are going to be my fish.

I fish models.

The advantage of those is of course
they're very simple to change we can alter

the material properties very easily by
changing out this bending surface as it

swims we can change the shape of the tail
by using a pair of scissors which is hard

to do on a real fish but of course
they're not very biologically realistic.

So I think somewhere by
taking a range of approaches.

Hopefully we'll learn enough
that by the time I retire.

We'll made some progress in trying
to understand a fish fish function

from a robotic point of view.

So let's begin with some whole
fish robotics very simple.

I'll just show some movies and show you
a little bit of data we have a couple

papers on this that you're
welcome to look at and

then we'll transition to the simpler
robotic models in a little bit.

So this is collaborative research
with Malcolm A Kyra's group at

Northwestern University
on the ghost nice fish

which is an amazing animal
found in the Amazon it

has an interesting electrical properties
like a sense and capabilities.

Let's see if we can get the movie
to play here should go full screen.

So look at the maneuvering capability
of this fish with this fan.

You can see it can go backward it can go
forward the movie should loop around so

you can see it from the beginning again
and basically does all of its locomotion

you little Petrel fin motions but
with this undulating ribbon like fin.

And there are quite a few actuators here
on the fish there almost a hundred thirty

individual fin rays each
FINRA has four muscles.

So we have a problem that's
complicated to understand.

It's pretty easy to see from all
the work that's been done and

just from intuition that when the fish
goes forward there's a traveling wave that

goes down the fin that goes backward.

It's pretty easy to understand
the fish wants to go backward.

There's a traveling
wave that goes forward.

But we noticed an interesting
thing about this fish and

you'll see it when the movie loops
around when it comes up in here.

You'll notice the body
goes up right there but

it goes up when it does that the fin
moves in a colliding wave pattern.

And we wanted to understand how that
was working a mechanic Lee and so

Malcolm the Khyber is a group of
Northwestern built the knife is robot

which is actually quite
engineering tour de force.

There's actually a picture of
it in my lab and the flow tank.

There we simplified to thirty
two actuators stacked end to end

we have various force measuring capability
of their light emitting diodes that give

us error codes of which there
were quite a few but we were able

to make this ribbon thin move quite well
and measure the flows that come off of it.

So here's an example looking from
below in the float tank in my lab

here is another view from the side.

We haven't tilted up because
the fish sometimes swim tilted so

we were able to alter
the angle of the of the fish.

And we can program in the way
of form the amplitude.

Most parameters we want
including a colliding wave.

So let's look at what
the colliding wave does.

So here is the robot we're doing a Ph D.

analysis so we're imaging with laser
light and look at the particle flows and

the ribbon thing is right here it is
generating a clotting wave coming in.

Towards the center.

And you'll see there's a cone shape
downward jet that's been formed

right here moving down till it hits
the bottom of the flow tank and

this is producing the force
that moves the NY fish up.

So when it heaves up makes
a colliding wave with its fan.

That's a result of
the momentum of the water that

is pushed down with that colliding when
we've looked at quite a few details of

this this just shows
the streamlines here on the right.

Illustrating that pattern.

So we've used this robot and.

A whole series of investigations of
it to begin to understand what a way

forms on this fin mean when we noticed
interesting behavior from the fish.

We're able to show that this colliding
wave produces an interesting

one hundred dynamic pattern that
hadn't been the noticed before.

Now we're also continuing to work on whole
fish robots in Jamestown Gora's group at

Drexel has been working with us doing
the primary design work to build a whole

sunfish that looks like
this with a rigid body but

with various sensor ports on the side.

Each thing here is actuated so that we
can move it in a biomimetic fashion.

Here are some views of
the modular component of this

here it is in the water
before it started to leak.

So we have a few issues working
with aquatic robots is challenging

when you have something like this but what
we hope to use this for is to begin to

look at multi fin coordination move
forward and backward motion of the robot

comparison to what fish can do and flows
that come off of individual feeds and

then interact with flows in the tail so
we've documented from our.

Work on fish that upstream fins modify

the flow hundred dynamically before
it encounters the downstream fans and

this has an important effect
on the dynamics of propulsion.

But now we have a model system we
can use to to look at that in a more

biomimetic fashion.

So this is sort of as
far as I'm concerned.

Before I retire this is my
ultimate biomimetic device.

But to get there.

We worked on individual fins and
so here is a robotic pectoral fan

that's attached to our apparatus
is not part of a biometric fish

with individual fin rays that are printed
so we can modify the flexibility

of those we can attach strings and
tendons and make it move back and forth.

So we can move this petrol fit in
a flow tank in a vertical laser

light she we can image the flow that comes
off of that and one of the things that we

did was we imitated a very interesting
biological observation we had for.

The fish's pectoral fin when it moves and
that's shown in this view of

the robot here on the right from behind
where you can see the fin cuts the curving

nature and as it comes down hopefully you
can see the flow going around both sides.

So there's that if you're used to the
flight literature there's a leading edge

vortex that's formed on insect wings but
the pectoral fins officials like this form

two leading edge board sees one on each
side of this curve confirmation that comes

around and our our research seems to
show that what that does is that enables

the fish to because you have opposite sign
board to city around the top and bottom of

the thing that minimizes the center of
mass motions as the fish goes forward.

It's not the words it's not like a penguin
where you have the wings flapping up and

down and every time the wings go down
the penguin goes up and vice versa and

you see them going like this through
the water the sun just moves very steadily

through the water and in part that's
because of the opposite sign vorticity

rotating around the two edges of that
pectoral fin now in the tail robot So this

is our sort of test platform of the caudal
fin of a fish this is the tail of a fish.

We have tendons going down
into the body and out and

we can move this tail
in different ways and

in the next slide you'll see some
different motion programs for the tail so

we can move the tail up and down in a flat
conformation we can roll it back and

forth a number of different ways or we can
produce this cupping motion that you saw

for the Petrel fan here
where the tail cuts up and

then turns around and
will cut down into the flow.

There is coming down into the flow and
we can measure the forces and

see what effect these different types of
motions have on the force pattern and

if we look at that what we found which
was I thought was quite interesting and

perhaps not entirely surprising to the
fluid dynamicists that this cup in motion

the red triangles if we plot of thrust
force against frequency becoming more.

Produces the highest thrust forces at
a given frequency once you get about

half a Hertz here so that when this
cupping motion a fish fins where

the surface doesn't move flat and it
doesn't move with a defined leading edge

it moves with two leading edges into
the flow is a very common phenomenon

in fish biomechanics we see it
in a lot of different fins and

it's something I think it's important for
robotic devices to be able to do so far.

Many of the whole fish robots have
plexiglass fins that are rigid.

So I think the ability to modulator the
confirmation the fan is very important.

So that's the by of the medic
robot side of things.

Let's now go the other extreme
to something really simple.

I like simple.

I'm getting older I like
the simpler the better for me.

And so we decided to build in
the lab a robotically controlled

flapping device that can measure
the forces on the flapping objects.

The flapping objects will swim in
a research letting flume and we can then

swap out very simple different materials
and study their swimming performance.

So hopefully this movie will play and
that movie will play out as

this one stopped but you can see what
we have here the looking from below.

We have a piece of plastic like like
this that were heaving back and

forth attached to its leading edge
only and you can see that here.

So here this will be flapping back and
forth like that and

it's making a nice wave form down
the body and it's actually swimming.

So the system we have uses
a research relating flow tank and

mounted up above it.

You'll see in the next slide as well as
a low friction air carriage system that

allows the foil to generate thrust and
swim and

then we tune the flow past it to bring
it back to its initial position so

even though it looks like we're holding
this in place upstream and downstream and

just allowing it to move like this.

In fact it's free to move upstream and
downstream to it just doesn't.

Because its thrust is matching the drag
incurred by the water moving past it.

There's water moving past
it generating thrust and

we've tuned the flow to match the drag so
that it's.

Swimming and
it's really swimming self-propelled speed.

And likewise here you see this apparatus.

Playing there's a rod holding its
flexible foils there's an A.T.F.

force torque sensor here so

we can measure the forces in the two
arcs and measure the cost of transport.

It turns out that these foil swim it
Reynolds numbers in St Paul numbers

which are very similar to swimming fish.

So we feel like we have a very reasonable
model here of a swimming fish and

here is our carriage system up above
on the air carriages with heaving pitch

Motors this diamond now it's attached
to a rigid foil original boom plate but

we can swim this and measure hundred
nomics measure kinematics measure

efficiencies do lots of things with it.

These are meant to imitate the undulate
Torrie bending patterns you see

from swimming fish years and
you know he was a trout.

So this bending flexing motion.

You see in swimming fish is what we're
looking to imitate with our flexible for

oil system.

And when you look at the midline
patterns here is that a clown my fish.

Here's the bluegill sunfish and
here is what I call a stiff boil.

Here's a flexible foil you can see we can
do pretty well at imitating the motion

of a fish with these flexible materials.

And we can measure the factual
stiffness which we've done so

we know exactly what the flexible
stiffness of that is.

So let's just do some
very simple experiments.

Let's look at the effect of altering
stiffness so here's an anesthetized fish

that's not very stiff and in fact it's
a weed match the Flexeril stiffness of.

Several of these plastics through the fish
so we we've got a biologically realistic

flexible fish and
then let's just alter the lane.

So experiment we can't do in biology we
can't just make the fish twice as long

keeping everything else constant.

If we alter the stiffness we see
a pattern that looks like this.

So this graph shows the swimming speed the
self-propelled swimming speed the speed

at which these flexible
plastics naturally would swim.

Plotted against the stiffness
of the material so

each point is a different material across
this graph their error bars in here but

they're they're hard to see they're
kind of hidden in the symbols.

And fish Flexeril stuff this
is right about in here.

The passive body of fish
lies right about here.

And what you can see is for

a particular way of moving the foils
if we look at the dark blue diamonds.

We see a peak right here and then a drop
off in swimming speed with stiffness.

But if we move it slightly differently.

We add some pitch to the foil so
instead of just moving it back and

forth like this.

We're going to add some rotation to it.

We can eliminate this peak and

have a long performance plateau that's
high appear that light blue squares.

And so fish should be able to alter
the performance their swimming

speed their performance
by changing the nature of

the way they activate different
segments of the body.

We know this to be true of course for
measuring things on swimming fish.

What these data show more quantitatively
is that by adding some pitch

moment which fish muscles can do to
the body segments to the swimming fish.

You can eliminate a performance detriment
that you get as you stiffen the body and

when fish swim faster they are stiffening
the body estimates now are they.

Steffen about two to three
times their passive body

stiffness as they swim faster.

So now let's do the simplest possible
experiment let's just make these

plastics longer.

So we're going to take a material that's
like this going to make it longer and

ask how well does it swim so this plot is
going to be self-propelled swimming speed

and centimeters per second
against length and centimeters.

If the foil is very small you don't expect
much thrust to Foyle's very long you're

going to get a lot of drag because we're
only moving it at the leading edge

somewhere in between.

Maybe we get a.

Peak.

So we would get a curve that
looks something like this.

This was my expectation.

As some of you have seen before.

This is what we actually found.

I have to tell you this is interesting
because on the monitors in my slides it's

red but here it's.

Black it.

That's interesting because I have
different color graphs about to appear so

you can see that we find very
non-linear patterns of change

in swimming speed with length
some lengths are very bad.

Some are better and this has to do with
the nature of the flow down the foil how

it's bending relative to
the flow past it and when

you change the stiffness of the mature so
we think of these as resin troughs and

peaks and they're pretty narrow they're
within three or four centimeters.

And when we look at the colors or.

What's exciting.

These are all very different colors.

But you can see that the different
materials from most gifted medium

to least.

I think you get the idea there
is a very different pattern of

swimming speed with Wayne.

That depends on the Flexeril stiffness so
I was pretty confused by this.

So I called Silas Alban here it applied
math in Georgia Tech and Silas made a nice

mathematical model of this where his Our
experiment and his model experiment and

model and then his experimental peaks and
troughs year which he was able to then

calculate with much more precision than we
could do in the in a laboratory setting

that the point here is you can see narrow
troughs and peaks resonant peaks and

troughs through this particular material
play again plotting swimming speed

against length that show that the system
is quite sensitive to length.

So as fish change their length
like all other things being equal.

They're going to be some significant
changes in swimming performance that

you might expect if they don't compensate.

In other ways I think is no evolutionary
biologist It makes me a little

nervous because quite a few claims have
been made about different populations

of fish that change their body size and
I think that shows you the.

The body size effect
particularly Wayne can be quite

quite complicated to understand.

It's a very good question.

It's a pretty recent result for us so we
have not actually looked at these in fish.

I think the alter of a fish

just in length is not that
probably that easy to do but look.

Actors.

Appear lake effect no no one has done
that yet but that would be up to us.

I think.

The frequency will change that plot
a fair amount and the it takes a while to

do the experiment so that I left a silence
to do in the computational domain to

affect frequency of the amplitude of
the heave motion can change things too and

whether you add pitch it's a very
big parameter space to explore and

you're absolutely right it does change
that and it can change the pattern

with stiffness so I think there's a lot
a lot to learn here about the fundamental

dynamics of aquatic swimming
stiffness versus swimming speed.

So I think I had a key issue still to
resolve with this if you think about how

we did these experiments.

We've got a very low mass foil here
this thing is fifteen grams or so

the carriage up above the controls this
and moves that is about four kilograms.

So the one thing we haven't done
a good job of imitating is the center

of mass motion of this while this oil
was swimming in the middle of the ocean.

It would be bending back and
forth it would also be moving upstream and

downstream slightly as it swims as it
generates thrust at different points in

the flapping cycle so there should be
some center of mass motion to this boil.

So in our effort to make
a more biologically realistic.

Well bought it.

System that imitates not just the bending
of a wave form of this with the flapping

foil but
also the center of mass motion and

you put the colors are really wild here.

You're not you're not seeing parts of it.

We decided to modify a robotic
apparatus and if you don't modify it.

This is what you see I'm going to show
you a movie here of a bending for

oil in a laser light with particles and

we're going to plot the force simultaneous
to those measurements to this images.

The force versus time and you can see
that we see force oscillations with

in a cycle which is what you would expect
to see with this set up as you have a very

heavy carriage this thing is pushing
against that heavy carriage.

It's pulling back against
a heavy carriage and

so you see these large force
oscillations through time.

But if this was a true freely swimming for

oil that should be near zero These
force oscillations would be nearly

instantaneously zero as the center of
mass is allowed to move back and forth.

So in order to fix that
we added a a motion to

the robotic flapping for the device that
imitates the center of mass motion and

I've exaggerated it in these movies so
you can see it play this first movie.

You can see the carriage here heaving
back and forth which is moving

the spoil back and forth but we've added a
linear motor here that's pushing back and

forth on the carriage at the same time.

So we're forcing.

The swimming for oil here to move back and
forth like this but also upstream and

downstream.

And you can see that here where we look
from below here is the bending for

oil again I greatly exaggerated this I
think this is a centimeter upstream and

downstream but
we're moving it upstream and

downstream at the same time
we're moving it back and forth.

Now we want to look at the parameter space
so let's look at changing the phase and

changing the amplitude
of this oscillation So

we're going to move it up backwards
forwards a lot then a little bit.

We're going to change the phase of
when we time the motion relative to

the heave side to side and you can see
a nice minimum here where this one's about

half a millimeter where we find we
minimize the force oscillations.

And so it's.
We apply this phase of a back to seventy

in about a half millimeter
upstream downstream amplitude and

look at the movie again this trace up top
shows you the linear motion is a plus or

minus about half a centimeter
half a millimeter and

on the bottom then we get a force trace
that's not exactly zero we're down to

around five million Newtons
which is not too bad.

Given everything else.

So what we feel we've done here is
to convert this for oil swimming in

the lab into what it would be like in the
middle of the ocean that is it's allowed

to have center of mass oscillations it's
allowed to flat flux back and forth.

So we feel we have a nice model for
a really swimming fish that includes.

Oscillations of the body in
the upstream downstream direction so

that begs the question what are fish do
in the upstream downstream direction.

What if in a live fish whether it's center
of mass oscillations or at least it's X.

direction upstream downstream
body oscillations which is what

they really aren't that I'm
going to measure here for you.

What the magnitude of those if you have a
twenty centimeter fish does its body move

up and down a centimeter
millimeter Micron what it was

was it moved and the way we've done that
too is to measure the center of mass or

the least the body position oscillations
I should say in the following way.

And people who have tried this in the past
have done things like trying to digitize

the nose of the fish here.

And you can see that that's not so
easy to do and

you can light it a little differently but
you have one pixel error up and

down stream and you get kind of a mess
when you differentiate that twice to look

at accelerations so what we've done is
use a pattern matching algorithm to natch

the pattern to track the pattern over
the center of mass location on a fish.

In different views and

track that through time and these are the
velocity vectors of the motion of this

portion of the fish at this moment in
time we can track that through time.

Take a mean if you just a much better out
estimate of body velocity through time and

here's a movie that shows
you what this looked.

Looks like here's a little sunfish you're
going to see yellow velocity vectors

you're tracking this pattern
of this region of the fish and

the graph on top shows you the tail
beating back and forth in the dark blue

and white is the acceleration
of that point on the fish

in the upstream downstream direction and
typically you see twice the frequency

the body accelerates forward and
back twice the frequency of the tail be.

So now we have a good way of
tracking this we can look

at the motion of the fish in this upstream
downstream direction and if you remember

from the flapping Foyle experiments
we found about half a millimeter

was the amplitude that is the peak
amplitude of the sine wave oscillation.

And when you look at a variety
of different swimming fish.

If you double that you get about
a millimeter this is the amplitude of

the oscillation in the X.

direction and you see for
different fish and varies a little bit.

It's around a millimeter the mathematical
amplitude be half millimeter

of the wave upstream and downstream so I
think we have a nice system down to begin

to explore what center of mass motions
of fish do what oscillations in body

posture occur during locomotion it's
a little embarrassing I think for

us in the aquatic realm because
terrestrial workers and

robotics the center of mass is
really important tracking it.

But we've ignored it completely in the
chronic realm but but not for much longer.

Hopefully we can continue to make
progress in this regard not want to end.

With some of our newer work using the
robotic flapping foil devised to attack

a problem that has been well recognized
in the bio bioengineering community and

that is the structure of sharkskin and
up connect this to the robotic foil

see the connection to the robotic
system in a moment but for many years.

Biologists certainly have
recognized that shark skin has

an interesting structure to it and bio
engineers and engineers have used this as

sort of generic inspiration for
making bumpy surface is that reduce drag.

There are quite a few papers sort of.

With the Sharks in bio inspiration
trying to look at the effects of

that on the surface drag
on the on on a particular.

Plate or coat the plate with
different materials but

to my knowledge all of that work
has been done on rigid oils

rigid platforms coated with material flow
is moved past it and the drag is measured.

But sharks course move.

That's one of their most important
biological characteristics.

So I thought we had an opportunity
with a robotic flapping foil device to

study the locomotive effect of
sharkskin while it's actually moving

what you see here is a picture of
a bonnet head shark with S.C.M.

images of the scales from
different parts of the body.

This is a very characteristic these are at
this image actually rotated ninety degrees

the scales don't point down they
point back these prongs point back.

We can look at zoomed in the scales near
the head of this sort of flat paving

like nature but the ones down
the body have these ridges and

here's a field of these and here's
a blowup of one of those and they're about

one hundred thirty microns in Dimension
them a very intricate structure and

I should say these are bone people think
of sharks as being Carnlough genus fishes.

But sharks have bone and here it is their
cut their skin is covered with bone.

And even better than that each one of
these is actually a little too just like

your two that it's got enamel it's got
Denton has a little pulp cavity inside it.

So this is where your two teeth
came from a long long time ago

but you can see that a very interesting
structure they've got these ridges would

stand up a Staal up surface.

It's inspired the speedo
wetsuits these are now banned but

the small ones are been sort
of inspired by shark skin.

There's a loose inspiration of
Ribot's these groove materials that

are used on wind turbines and other.

The Americas cop I think from one
thousand nine hundred seventy had

reports on the hall and
did really well when they banned it.

Because it went faster.

There's general shark inspiration for
these these structures.

So what we thought we could do was
to test the effect of shark skin and

some of the biomimetic material so we have
we made our own material I have a sample

here if you want to see it afterwards and
we have some of the speedo fast into

wetsuit material you can see in about
ten seconds it's not very biomimetic but

I'll show you a picture in a moment and we
can study sharkskin we have this flapping

foil apparatus what we can do is get some
nice freshly dead sharks we can chop

out pieces we can clean off all the skin
underneath and make the skin strips and

we can glue them to each other and
make a membrane a sandwich.

Of sharkskin.

So that each part of the outer surface.

Has the surface structure of the skin
has the dentical structure of sharkskin.

And we can make these into for

oil is that we can attach to the apparatus
so we can make a rigid foil so

here's a rigid plate aluminum plate
covered with sharkskin on both sides and

here is a flexible foil that we've made
out of the two layers of shark skin and

what you can see down here but the dark
outer area is the skin surface and

then the little white area in the middle
like an ice cream sandwich is the skin

that's underneath that
we glued to each other.

And these will bend.

When they swim and we imaged just
downstream of this to show that

you know what the scales and
the scales are called scales or dentical.

Look like they're.

So we were able to make a membrane that
approximated the curvature that we see in

actual swimming sharks so we swam sharks
in the lab we measured the curvature of

the body down here in the mid tail ish
region we compared that to the curvature

that we see here on our shark skit
membranes of this is a minute by layer

membrane of actual shark skin and
we were able to swim these so.

This movie on the bottom here shows.

US heaving this membrane of
sharkskin back and forth plus or

minus two centimeters with twenty degree
pitch between one and three Hurst we moved

it in different ways and you can see this
membrane is swimming in again it's in our

freely swimming apparatus so
this membrane of shark skin is swimming.

And it's self propelled speed and

whatever its natural speed
would be given this motion.

Given this input heave motion
like you want to treadmill

you set the speed at two miles an hour
you're walking to match that speed.

If you increase the speed
you have to walk faster

here if we increase the speed
the sharkskin wouldn't be able to keep up.

It would move downstream so
we we matched the flow to its thrust.

Here is the speedo material.

This is they're fast into material which
they don't make wetsuits out of anymore.

This is the underside.

And that's the biomimetic side so

I hope that doesn't look like
sharkskin to anybody here.

And when you cross section it you'll
see basically what they've done is

they stamped dents in the surface and the
dents look like it should make you go fast

on the pattern the top of the deaths are
quite widely spaced about a millimeter and

a half apart.

We didn't think this would do much.

So we made membranes but
we made membranes out of this mask and

fabric with the biomimetic that surface
on the outside of the flap them back and

forth and then we made other membranes
where the bottom the Minix surfaces

on the inside and we flopped back and

forth it's important to keep the Flexeril
stiffness of the material constant.

Buzz want to change the surface.

That's in contact with
the propulsive water.

And what we found were
things like this for

the speedo material see a grandmother
graph that looks like this.

This is swimming speed on this axis.

And these are different ways we moved
the foil different motion program so

through these two bars you can
see we move to two hurt plus or

minus two centimeters with no pitch for
this one.

We went three hurts plus or

minus two centimeters twenty
degree pitch and you can see here

there was no significant difference
in swimming speed they both swim

almost twenty centimeters per second with
the biomimetic surface on the outside.

Compared to on the inside here
it swam slightly faster but

here it swam quite a bit slower so he had
this motion program the bottom of a slow

down the foil so on average we concluded
the speedo surface doesn't actually

do anything now the whole suit does
things the whole suit reduces your drag.

It's not because of the surface
we can talk more about that.

Figuring out the control for shark
skin was not so easy because we needed

to swim the membrane with the gun tickles
and then I wanted to remove the Democrats.

But I didn't want to alter
the Flexeril stiffness of the skin by

chemically treating it or

trying to pick out with forceps each
individual hundred micron scale.

So we took the sanding it off so
we wet sanding down the surface and

you do see a number of little nubs
present here there's no doubt about it.

These are bedded in the skin.

It's very hard to remove all the knobs.

But certainly that looks
a lot different than this.

So let's compare the swimming
speed of membrane like

flexible foils made of normal Sarsi and
after we sand off the surface and

what we found was in every case
the normal sharkskin swims faster when

it's part of a flexible foil when
it's allowed to been back and forth.

Here's a case where intact sharks
can swim faster than when we

sand it off when we moved it in this way.

It's when fast and when we send it off
when we moved it like this when faster

than when we sent it off the average
benefit was about thirteen percent.

And interestingly this was not the case
for the SAARC starts to get attached

to a rigid foil on a rigid foil sanding

the surface actually reduce increase the
swimming speed increase the swimming speed

I think because the wetted surface area
is down but when a flexible membrane

bent like a shark swims its wings
faster with the Democrats attacked.

There's an interesting
other part to this story.

Now we want to just look at the flow
over the shark skin membrane so

here's a shark skin membrane.

Here's a laser light sheet
we're going to image the flow

as this membranes move back and
forth in by our floppy for all device.

And what we can see here

is my turn the lights off a little
more that's it that's OK So

here's our sharkskin membrane we have
laser light coming in from the bottom.

It's making a shadow here flow freestream
flow is going down left to right and

you can see the foil bending here and

you see this large leading edge vortex
that's formed on this foil as it moves up.

There's one on the in the shadow here
on the other side as it moves down.

It's about to move up
here we go builds and

builds a build it's actually really hard.

It's a couple of centimeters and
then it's destroyed and

peels off when the foil
moves back the other way.

Now this type of motion is very similar to
what happens on a live shark tale the back

and body that's been demonstrated their
separation leading edge border sees

have been certainly suspected in
this membrane now we can see this

leading edge more attacks and you can see
how close it is to the surface the flow on

the surface here is actually moving in the
opposite direction as this big tech spins

up what we noticed was that
the position of that vortex changes

when you sand down the surface
tentacles we can see that here.

This last data slide.

Here's the sharkskin foil.

Here's the vortex.

Here's the sharkskin foil where we've
sanded off the surface bony genitals and

the vortex moves further out and
this these plots to show that that peak

vorticity is about a centimeter further
away from the foil when the sand

off the dental is to when it's intact and
this suggests to me that the shark skin

Demichelis people have been thinking them
a drag reducers which they may well be but

that they might also be enhancing leading
edge suction that is the suction force on

the leading edge of the tail by
promoting enhancement of that vortex.

On the surface of the tail that vortex.

The closer it is to the surface
the stronger the leading edge section and

you can get a nice That has been a thrust
by the dentals not just reduction of

drag that's a hypothesis is that
the devils might be doing both things.

So let me wrap up and say that I think
that as an evolutionary biologist someone

who kind of wound their way through
a Securitas route into engineering and

bioengineering that really
it's been a tremendous benefit

because it lets us do things we just
can't do just measuring the forces on

really swimming fish has been
a pain in the neck over the years.

You know there are things we can
do we can quantify the way we can

look at momentum flux in a way but
it gets of life issues this is just hard.

We can track body accelerations that's
something that's we're doing more of

that's not so easy to do and
it takes to a lot of time.

So a robot we can attach our force
transducer it's a big benefit I'm

very jealous of the terrestrial
locomotion people because it's

just easier to measure forces but robotic
devices for all of us were great and

I think the biomimetic devices are very
valuable but they are hard to build takes

a few years to build them and
I fish robot took a few years to build.

Take us about a year and a half so far to
get the whole this robot with all of its

pins actuated and not leaking and
it's just it's a challenge.

So I think it's a very
valuable approach but

I think it helps to have another kind
of approach at the same time and

I think that the simple robotic models
just the simple flexing for oils.

Even very flexible things like
this swim actually quite well.

And we can study the effect of stiffness
we can study effectively we can swap out

for sharkskin we can study lots of things
about the functional design of fish is

that a very hard to do in a in a lot
of fish and so I like very much.

The the approach and I think that in
combination with studies of life fish.

I think we have an opportunity to
make some real advances in our

understanding of aquatic.

Olson and
of the engineering of human devices to

allow us to do things in the water
also thank you very much.

Happy to answer questions
an hour later yes.

The it's a very good question.

The Danica's themselves are not active but
it's been proposed by a couple

of groups that the bending of
the skin enables them to bristle and

flare out and back and
that suggests you know maybe they're

pumping water out into the boundary
layer as they're bending back and for

them there could be on god knows
a number of things going on and

nobody really knows yet nobody has
imaged what the Democrats are doing.

At you know.

Half cent and
half centimeter square field of view.

As they're swimming but that's certainly
something we're interested in doing.

Or they're very they're very three day.

Absolutely.

So normal fish scales are more
like flattened layers stacked

like this shark scales stand up there.

They're like a tooth and they're cups.

So we've actually imaged using
a microscope to reconstruct

in three dimensions the stark scale who's
making a biomimetic version of this is

actually not so easy because
there are you're at the hundred

micron scale there are features
you'd like a thirty Micron scale.

There are three dimensions so it's yeah.

So they sit up kind of like this their
little stalks then they flare out and

then the next one is like that.

Nobody's ever quantified.

Are they randomly spaced Is there a ray of
these and that that's something we could

test with a biomimetic skin but sharks
do lose them and they can regrow them.

Yes.

Well it's an interesting question because
the catfish actually are part of a group

that is not the spiny thing fishes but
you're right that they have those spines

and then the spine even fishes that
I mentioned earlier have rigid spines

that actually are not able to move side
to side they can only go up and down and

that seems to be something to fish use for
maneuvering also as well as for protection

when they're executing it turned out put
that up like an instantaneous heel being

erected in this case and it's difficult
side force as opposed by the fluid.

So I think in case of catfish
they have a pectoral spines

which you know no one's really looked at
their their use in locomotion but they're

certainly used defensively there certainly
is a wedge themselves and crevices.

And there are some Capra
species actually rub them.

The basic against their body to make
sounds and communicate with each other so.

The tapestry interesting special
case I would say but for

other fish they would have the spiny fin.

The ribs or spine and I think does play an
important role in locomotion hasn't been

studied all that much.

Become over here does in that.

They do so.

Yes So when the federal thing comes
out that that the amount of cupping in

there will change with speed it
will it will change the speed and.

To Remember it could be.

It's a little tricky because
as the fish go faster.

I think it is it's more cup with speed but
the problem with the fish I showed

you is as a fish like if you know think of
like locomotion horses horses that gait

they go from a walk to a trot to a gallop
fish have gates to actually so when

they swim in a slower speed they're using
the petrol fin with this cupping motion.

But they're reaches a point actually
which they can't generate enough thrust

with their pectoral fins as they start to
use their tail and their dorsal and their.

Offensive though were crude other fins and

then as they go even faster they
start to really bend or body.

Not all fish have the same gates but many
fish have that pattern that I described.

So the cupping increases but then only to
a point at which point they sort of stop

using their pectoral fin so much and
use other muscles for the back the body.

Oil

Well there's that's a very
difficult question and

I really I can't give you a good answer.

I think because the surface
is a fish the fins

are very flexible and
they're subject active control but

the fluid forces are still large when
you see any given amount of bending you

actually don't always know what
percentage of that is due to active

control by the fish and what percentage
of is due to impose forces by the fluid.

So the the best way we've been able to
work on that is record from the muscles so

if you see the fluid bending the fin
back what might look passive but

all the muscles that would pull the fin
forward are active in the fish.

We conclude that the pattern you're
seeing is a combination of active effort

by the fish to stiffen the fin and loading
by the fluid on the thin to to bend

it back so I think most for
most moderate to fast winning and fish.

It's a combination I think the fish
are clearly actively tuning their

bodies and their fins.

Absolutely.

But they're also subject to
substrata fluid loading that is

going to also impact of the bending
pattern that you actually see.

We can maybe we can talk about so
it's a it's a tricky and

complicated non-trivial question.

Two and a half.

Right.

So we've worked on with like
a drop the feel is group at M.I.T.

a dead fish will actually swim in of why.

Tech street and there is although
they're not so dead there Steph.

But if they're flexible dead.

They'll swim in a vortex street and
you get a Carmen St and

the fish will can tune their body to
the width of the street and the oncoming

boards to sexually generate thrust and
move upstream against an oncoming flow.

So a purely passive fish can generate
thrust that's one wrinkle to your to your

question but I think in most
fish swimming most of the time

the pop pattern of body bending you see is
a combination of active and passive and

it just makes it tricky to UN
cover the physics of the system.

You mean different
amplitudes longest later.

OK so.

So while your question is.

OK.

A little bit.

Yes I think that's true.

So I would say that's true and for very
flexible foils it's it's more obvious for

a very flexible material where you can
see many wavelengths that you can.

Use me just shorten it up and

see that you're short of the number
of waves even double the length of

each wave can actually stay that
constant when you get to a stiffer

material the wavelength can change as lane
changes too because the fluid is pushing

on the on the on the material which
is still somewhat flexible and so

we haven't tried to isolate that I'd say
it's a cop my my my not so good answer.

It's got the complex interaction and
what's one reason we like the swimming

speed the self-propelled swimming speed is
our performance metric because that kind

of integrates all of that into one
just ask how well does it swim.

Where.
That becomes very important as asking Well

then why is it swimming so fast and

I think we don't have
a full answer to that yet.

I would say.

Time.