Jon Kuniholm, "Good Design by Design" | Talks at Google

[Jon Kuniholm - Good Design by Design
May 2nd, 2014[

MICHAEL WEISS-MALIK:
Hi, everybody.

Hear me OK?

OK.

So I think we're
going to get going.

My name's Michael Weiss-Malik.

Our guest today,
Jonathan Kuniholm.

I met him in, we
think, 2007 as far

as we can piece back together
at a SciFoo conference.

And he had some pretty
cool tech that he

was demoing on some open
source robotics stuff

related to prosthetics.

And I was impressed
enough that I said, hey,

you should come and give a
talk at Google some time.

And he took that to heart.

And about four or
five years later,

emailed me and said, hey, you
offer that talk at Google.

Now would be a great time.

So we're hosting him today.

Jonathan is the president and
founder of the Open Prosthetics

Project.

He's also a founder of
StumpworX, a startup that

focuses on prosthetic
technology.

He's also presidential appointee
to the National Council

on Disability.

And the stuff he has
to say is pretty cool.

So please put your
hands together

and help me welcome Jon.

(Applause)

JON KUNIHOLM: Thanks
very much, Michael.

Let me first just offer
a disclaimer with respect

to my government job.

They encourage us to do this.

That everything that
I'm going to say today

is-- are my personal
views and not

reflective of any position
of the government.

So what I would like
to talk about today

is, how we can design good
design to solve problems

that society has, for
the most part, neglected.

And I came across
one of those problems

personally after I
lost my arm in 2005

and discovered that
prosthetic arms were

an orphan medical device.

And in thinking
a little bit more

about why prosthetic arms lagged
so far behind other technology

that we use every day,
I started to realize

that prosthetic arms and
orphan medical devices

are part of a larger
group of those problems

that society has
tended to neglect

which can be solved by
something that we're beginning

to call public interest design.

And the question
that I'd really like

to talk about today--
and I would actually

like it to be the
beginning of the discussion

because it's something
that, by no means,

I claim to have
even begun to solve.

--is, how can we marshal all of
the tools at our disposal to,

trying to better
solve those problems.

Because very much now I believe
that most of those problems

are solved in kind
of a haphazard way.

Government-funded projects,
philanthropy, side

projects from industry,
you name the way

that people happen upon these
issues and try to solve them.

But in general, you can
be sure that the resources

and attention that we
devote toward solving

these underserved
needs are going

to lag far behind
those problems which

are very obvious from every
other standpoint need solving,

very profitable things.

There was no question
that cellphone technology

was going to improve over the
last 10 years, for example.

So in the summer of 2005, I
took leave from graduate school

at Duke University.

And I was deployed as a Marine
to Iraq in Anbar Province.

And I was the platoon commander
for an engineer platoon

of 50 Marines.

And we were doing
mostly what everybody

was doing over there which
is sustainment and stability

operations, patrolling, guarding
convoys, that sort of thing.

On New Year's today
of 2005, I was

on a foot patrol
that was ambushed

by improvised explosive device.

And the blast took off
most of my forearm.

And I found myself back
in the States and learning

about prosthetic arm technology.

I managed to get
myself back to school

and get involved with a
research project sponsored

by DARPA called the
Revolutionizing Prosthetics

2009 Project.

It was one of two that DARPA
was funding which is really

the first serious prosthetic
arm research effort that

at occurred in the United
States since-- there was

a small one in the '70s but
really since World War II.

And the goal was a
really ambitious one.

The goal was to
create in four years

in 2007 an arm using
commercially-available

technology that could go to
market in that year, in 2007.

And then, more
ambitiously, to create

an arm that had more
degrees of freedom,

was fully neurally integrated.

So nearly the same articulation
as a native human arm

and full strength.

And an incredibly
ambitious project.

What I think is
important to understand,

there's been an enormous-- and
this picture is up here not

too brag that I
was on "60 Minutes"

but to show the
kind of attention

that these projects got.

For whatever reason,
prosthetic arms

tend to really capture
the popular imagination.

People love robotics.

They love thinking about the
barriers, the singularity

and the barriers between man
and machine disappearing.

And it becomes a vehicle for
all kinds of philosophical

and science fiction interest.

I think the takeaway
about these projects which

is important to remember
is, first of all,

that we-- despite the
amazing things that both

of these research
efforts accomplished,

we still have not, 10
years, more than 10 years

since the war began Afghanistan,
actually pushing 15 now

--still do not have any
commercially-available device

that has resulted from any
of this government-funded

research.

We're still waiting
for the first one.

It has remained, as always,
just around the corner.

And I'm told that we're waiting
on FDA approval for the DECA RP

'07 project to
receive FDA approval.

And it has been in
some clinical trials.

But we still haven't
seen a device.

The other one is-- I think
that's important to understand.

It's my belief that
we have a media bias.

It's not a political
conservative media bias.

But it's one towards
entertainment.

And so there-- the word
bionic is used a lot.

The words dexterous
and manipulation

are also used a lot which
have very specific meanings

in robotics.

And there are some
caveats that I could give.

You could call a trigger
grip, for example,

where one part of
the hand is grasping

a handle of a cordless
drill, and the finger

is pulling the trigger.

That tends toward a
sort of manipulation.

It's at lease a compound grasp
where one part of your hand

is doing one thing
and one doing another.

But of course, the slide that
I'm showing here right now

gives you some insight into what
the human hands are capable of.

This pair of hands
is doing a card trick

where four cuts of the
deck are being controlled

by different parts
of the two hands.

And it's a dynamic movement.

And so I guarantee
you that there

isn't a robotic hand
of any kind that

could do these card
tricks right now.

But based on the popular press
presentation of all this,

you would think that it can.

You would think that
these problems are solved.

So while I think it's important
to acknowledge the strides that

were made by these
research projects,

it's also very
important to understand

that we are not there yet.

And we do not have
bionic people.

And these hands are,
for most part right now,

I think that it would be fair
to say, at least in terms

of prosthetic control, that even
these highly articulated hands

are really only
capable of grasping.

And the difference between
grasping and manipulation,

you can use a Rubik's
cube to illustrate.

Manipulation is a
speed cuber solving

a Rubik's cube in a few minutes.

And grasping is not dropping it.

And grasping a Rubik's
cube is actually something

that I can do with
the hook that I

wear when I do wear a prosthetic
arm, which is the Dorrance

5x named after the guy
who patented it in 1912.

And you can see that since
then, it's evolved a little bit.

There's some rubber grip on
the fingers, a cigarette notch,

and I think it's
called a pen notch now.

But this remains, despite
everything that's happened,

the most used prosthetic
terminal device in America.

And were you to go to the
Myoelectric Conference

in New Brunswick in
Canada this August,

it would be a fair bet to make
that the majority of amputees

attending the conference, at
least the ones who weren't paid

to be there by a
prosthetic company,

would be wearing
one of these devices

and not any of the
robotic prosthetic hands

that are available.

The other thing that's
important to note about this

is that there's another
choice which actually beats

all of them which is the
one that I'm wearing today

which is nothing.

And about half of
all arm amputees

choose most times not to
wear a prosthesis at all.

And it's mostly because
of weight and comfort

in suspension.

And I'm going to talk about
that a little bit later.

So now, why is all of that?

This is by no means an
attack on any the folks

who were generous enough
to fund that project

and try to make some progress.

The real problem is
that there simply

isn't a market for these things.

The market for powered
arms is just barely noise

compared to these
other large markets.

And even though the government
has accomplished quite a bit

in some of these
research projects, what

we call the valley of death
which is the bridge that one

of these products must cross
in order to get from the lab

into becoming a clinical
product, is really significant.

And without a market to drive
further funding, that gap which

is estimated by
people in the industry

to be something like
90% of the R&D remaining

on a medical product once
a proof of principle device

is finished, there's just simply
nowhere for those resources

to come from and make it happen.

And so correspondingly,
the amount

of effort that's actually put
into solving this problem is

also very small, compared
to-- I cherry-picked

a few really large
government-sponsored

engineering projects.

Some of which have been compared
to the DARPA prosthetics

project.

This is all in 2006 dollars.

The Manhattan Project comes
in at over $20 billion,

the moon race over $120 billion,
and just the R&D for the Joint

Strike Fighter at
over $40 billion.

And so the $100
million, well more than

anybody had spent on prosthetic
arms in half a century.

The DARPA spend is basically
just noise compared to this.

And I would argue that the
task that they set out to solve

rivals some of these in
terms of the engineering

effort required.

It's a small project
instead of a large one.

But nevertheless, an enormous
portion of our brains,

for example, is devoted to
the control of our hands

in manipulations.

I don't know if
you all have ever

seen-- I don't have a
slide of it here here,

but the homunculus that neural
researchers use to represent,

they take a human figure and
adjust him proportionally,

based on the number
of nerve endings

in some studies of the brain.

So this little figure has
gigantic lips, gigantic hands,

and, depending on whether you're
looking at the censored version

or not, gigantic
genitalia, reflecting

how much of our
brains are devoted

to control these various
parts of our body.

And this is true both in
terms of motor control

and sensory control.

I works both ways.

And as you can see from that
slide of the card trick,

feedback, sensory
feedback is certainly

a huge part of our hands
being able to do what we do.

So what we have is a
three-way market failure

to solve this problem- Industry,
government, and academia.

All doing their parts, but all
unable to solve this problem.

And so -- so

But of course, it's
completely understandable.

And that's why we
call a prosthetic arms

an orphan device.

The orphan drug law
that was passed in 1983

defined patients in need of a
drug as being medical orphans

if their patient population
was less than 250,000.

So prosthetic arms
in the United States

are about, a couple
years ago, 43,000

so fewer than 50,000 patients
in the United States.

So that's 1/6 of the maximum
threshold for being an orphan.

Course, the orphan drug law
covers pharmaceuticals and not

medical devices.

So we don't have a companion
regulation or law or anything

that covers the creation
of medical devices.

And another peculiarity,
of course, of that

is, how we pay for medical
devices in the United States.

And I will get to that
a little bit later.

So as I mentioned before, all
of these medical orphans--

and the NIH-- and
this includes all

of the drug, the orphan drug
patient populations and others.

The NIH lists something like
10,000 orphaned conditions

on its website.

And because I happened to be
interested in prosthetic arms,

I searched for limb
absence, arm absence.

And so my particular
orphan condition

is actually not one of the ones
that's listed on that site.

So I would say that
there are probably more.

And so the question is-- and
this gets back to the graphs

that I showed before.

Because there's obviously
no market for it,

it doesn't necessarily
make sense.

I wouldn't argue, for example,
that it's a travesty that DARPA

didn't invest $100 billion into
solving this prosthetic arm

problem because the
same could be said

for any of these other orphans.

We simply can't
afford to do that

from a public health standpoint.

It's just not an effective
way to spend our resources.

So nevertheless,
feeling like this

is a problem that we'd
like for society to solve

and that it's not the only
one, this is the question.

How can we attacked
these problems

and solve some of them?

And it's clear that
we probably need

to think about them a little bit
different way than we are now.

My friend Chuck
Messer described this

as using your geek
powers for good.

The Cooper-Hewitt Design
Museum, in referencing

the 90% of the world
that doesn't benefit

from our consumer culture,
they had a design exhibit

several years ago called Design
for the Other 90% in reference

to that.

And there's another movement
here in the Bay Area.

For example, you have
a designer and engineer

named Ralf Hotchkiss
who makes a wheelchair.

Whirlwind Wheelchair
is their organization.

And they make appropriate
technology wheelchairs

that have larger balloon
tires on the front

to deal with uneven terrain
and broken sidewalks

that you're more
likely to encounter,

if there are sidewalks all, in
a lot of the developing world.

And they have a cool
model where they

try to set up factories as
a profitable businesses,

a social enterprise, to
manufacture this wheelchair

design in the area where
they might be used.

And this in contrast
to sometimes

what gets done in the
wheelchair space, for example,

is to send a kind of
rickety hospital wheelchair

over to that environment
where at last,

like a lot of medical
technology over there,

about 10 minutes
before it breaks.

And then, somebody has trouble
finding a part to fix it,

and it ends up getting junked.

So there's-- whereas I'm arguing
that prosthetic arm users

in the United States are
an underserved population,

they're not
disadvantaged, certainly,

to the same degree that patients
in the developing world are.

But all of these
patients are underserved

in some way or another.

So a good phrase
that's been used

to describe all of these
problems and the design

that we can use to
try to attack them

is public interest design,
design in the public interest.

And so really what
I'd like to talk

about is, what are
the ways besides-- I

will start off with
government-sponsored research

because that's a lot of
what's happening currently.

But what is our
whole quiver of tools

that we might use to try to
solve some of these problems?

And-- yeah.

Sorry about that.

So what are the
collection of tools

that we could use to try to
better solve these problems?

And I, like lots of
people do, I think

I'll probably come down on there
being some combination of all

of these that are necessary.

But just to brush over
them all, to begin

with government-sponsored
research and commercialization,

then DIY and the maker culture,
potentially an orphan device

law to service as a companion
to the orphan drug law

that we've had in this
country since 1983.

And then, what other
regulatory or policy changes

might be made under
our existing structure.

But by program managers
who fund, for example, some

of this government-sponsored
research, what things might

they do to try to raise the
impact of the research funding

that they're
currently giving out.

And then secondly, what can
we do in the private sector?

Are there ways that folks
in the private sector

might be able to
contribute in ways

that don't impact
their own bottom line?

And they can
potentially even have

a win-win in serving
these underserved folks.

So to start off with,
these are the three--

these are icons that represent
the three major engineering

research efforts that
I showed the funding

levels for- the moon
shot, at the atomic bomb,

and the Joint Strike Fighter.

And it's certainly clear
that there are some problems

that government has
to solve if they're

going to get it solved for us.

While we do have SpaceX in
a commercial space effort

right now, it only exists
because the government

is choosing to create
that capability.

And the funding's coming
from the government.

So that's an example, actually,
of spreading this stuff

around a little bit.

But it was clear that
sending a man to the moon

or building the bomb,
however you feel about

that having done, these
were certainly projects that

would never have happened
had not one of the largest

and wealthiest
governments in the world

decided that they were
going to make it happen.

And they happen
because, in at least two

in these cases, the first
two that I mentioned,

the results were non-negotiable.

We were going to
get to the moon,

no matter how much it costs
and how dangerous it was.

And those who were
in charge of the bomb

viewed it as a response
to an existential threat.

And that failure
was not an option.

It's probably-- the
government probably

would have spent anything
to achieve these ends.

Now, a lot of times,
some these very programs

and the other basic
science research

that the government
does are cited

as having intangible
benefits that spin off,

that we would be without were
it not for these other efforts.

And so Tang, Teflon,
and Velcro are three.

And there's actually
a myth-busting page

on NASA's website that I
have up there on the slide.

Says that, in fact, none
of these three products

were created by NASA
or NASA-funded research

although the space program did
take advantage of all of those.

So while they used Tang,
Teflon, and Velcro,

they did not create it.

And I would argue that even
if they had, building a bomb,

sending somebody to the moon,
or building a high-tech fighter

jet are not the most efficient
ways to create things

like Tang, Teflon, and Velcro.

So if what you want
is a prosthetic arm

or to put somebody on the moon
or to build a fighter jet,

that is what you do.

And this really comes
down to a difference

between engineering
and basic science.

So the point is that
whatever efforts there

are that are devoted to
solving this problem,

it really needs to be dedicated
to solving the problem.

And there needs to be
some agreement on actually

what the problem is.

A significant component
of at least the four year

DARPA effort was based on the
creation of a neural brain

machine interface.

And it is-- well,
in the long term,

it's certainly true that
that represents potentially

a holy grail of prosthetic arms.

It's not clear that that is
a necessary component of it,

particularly given the state
of the technology that most

people are using every day.

So then-- so I was-- at the
time when I met Michael in 2007,

we had a version of
this hardware board

that we were showing
Google SciFoo.

And this is-- Tim Hanson
is now post-doc at UCSF.

And he created, when he
was at the Nicolelis Lab

at Duke University, the
signal processing board which

is capable of processing 16
EMG, electromyogram, signals

from the surface
of the skin that

can detect muscle movements.

And we designed it to have as
many interfaces as possible.

Because are our goal
in this DIY space

was to try to encourage
experimentation

and to try to benefit
from the maker

culture and the huge
video game industry in --

And hope that some of
those resources and efforts

and interests could rub off
and make some improvements

in prosthetic arms.

What we learned is the
downside to DIY maker

and open source culture was
that, these projects sometimes

require-- well first, they
can't be done completely

without resources.

And then secondly,
there's activation energy

to any kind of
collaborative effort.

Which if you don't
cross, you have

a lot of difficulty
making anything happen.

So this project is still alive.

I looked.

There were commits this week.

But they're primarily-- but all
of those, as far as I'm aware,

are coming from the two
labs at UCSF and Duke

that are using the hardware.

So the schematics-- all
this is open source and GPL

to the extent that it
actually applies to all of it.

The schematics are up there,
all of the masks, the firmware

for the board,
everything is there.

But as far as I'm aware, nobody
has actually made one of these,

except for those two
labs that are using

and are active contributors.

And Tim made a pretty
complicated board.

It's like six layers and has
these tiny little vias and some

04 components on it.

We had some-- we made
this thing in a toaster

and had some horrible moments
where somebody breathed on it

or-- And then, you can't
tell what any of those things

are anymore, et cetera.

So as far as I know, nobody from
the larger DIY or maker culture

has participated
in this project.

Now, there's also-- there's
another part of the project.

We have functioning pattern
recognition software in MATLAB

which is not open source.

And again, this is
another barrier.

In order to use the open source
software that we have up there,

you need to have
access to something

like $20,000 worth
of MATLAB tool boxes

which you're only going to have
if you're either an industry

or you're in academia somewhere.

So these, again, are
some of the barriers.

But beyond that,
this hasn't really

had the impact
that we really hope

that it would have
when we started.

Although I'm certainly
proud we tried,

and it's a pretty cool
piece of hardware.

Here's another one.

We had a little project on
the open prosthetics project

to recreate a
prosthetic hook that

was designed in the '20s which
has been out of production

for something like
35 or 40 years.

And most of the guys,
and it is mostly

guys, who use this hook
are now 80, 90 years old.

And I live in the Midwest.

I think there are some
demographic particular areas

about where this
thing was prescribed

or what people
wanted to use it for.

And I still get calls from
people who find-- very few,

it's a trickle.

--who find this thing
on the web and want one.

And you can see here, this
is a picture-- somebody,

Bre Pettis put it
in the Thingiverse.

And a bunch of
people-- I got bunches

of people who want to volunteer.

They were offering to print
as many of these things

as they could on there first
generation MakerBots, which

is where I think this
printed hook came from,

to help the project.

But unfortunately, even
the latest generation

MakerBot 2-- which which
I have and really enjoy.

--it doesn't print
in a material that's

appropriate for actual use,
at least in this design.

This is made to be a
metal, a metal hook.

And you can see that the
resolution is really not

fantastic there.

So and what we discovered,
that one in the lower right

was printed using
a selective laser

centering rapid
prototyping process.

And we had some tolerance
problems with that

until the service bureau
actually printed the thing out

on a plate and then
machined off the plate.

The two jaws of the hook
did not fit together.

So we had something like more
than 50,000ths off on that

thing.

and.

I bring this up
just to point out

the limitations of
some of these things.

And you see this--
just a loop back

for a moment to the
breathless enthusiasm

of the press for a
lot of these things.

I can't tell you the
number of stories

that I've seen in
the last two years

about a high school
student who has 3D printed

a prosthetic hand that is
supposedly better than, you

pick the price point, of
some commercially-available

prosthetic hand.

And when you-- there's
really no scrutiny

at all that's applied
to these claims.

Very often, there are
completely externally-powered.

Like motors with strings or
maybe they're not even powered.

In any case, nobody really
closely examines the claim

that some kid has bested
a $40,000 prosthetic

hand in his bedroom
with a RepRap machine.

And while I consider
myself to be a part of

and fully embrace
the maker culture,

I think we also need to be
very realistic about what we've

actually accomplished
and try not

to run victory laps before
we've actually done it.

And I think that this probably
is one of those cases.

Now, the happy
ending to this story

is that there actually is
a prosthetic company called

ToughWare that re-engineer--
they re-engineered this

and fixed some of the
problems that we had actually

identified with the
original design.

I think you could
actually see it

in the-- that's an original hook
from many decades ago up there.

And you can see the
places it has been welded.

And so this company fixed
some of the problems

and where these
things were breaking.

And I do believe that
you can actually now

purchase a version of this
hook commercially again.

So I consider that
to be a success even

if we weren't the
ones who did it.

So lastly-- and there is
some bleed across here.

Michael did reference the
company that I've started.

I also decided to
back up a little bit

and solve one of the problems
about prosthetic arms that

may be, in fact, the most
vexing and least address,

which is how we attach
them to people's bodies.

This is traditionally done
with a hard carbon fiber

or fiberglass socket,
a rubber liner,

and maybe a plastic
socket there.

And I would liken a
lot of these designs

to wearing wooden
shoes and rubber socks.

And it does pretty horrible
things to your skin

when there's moisture
and friction involved.

And heat, moisture, weight,
and perceived weight

is really related to
all of these things

because people will tolerate
hanging less off of your body

when it doesn't really fit well.

It feels like it's heavier
if it's uncomfortable.

So we decided that
the way to solve this

was to go back to what
prosthetic arms used

to be like in the
'40s before the VA did

all the research that created
these composite sockets.

And consider making prosthetics
sockets more like shoes,

like athletic shoes.

Prosthetics sockets used
to be made of leather.

And I've been told
by old guys who

wear leather
sockets that they're

the most comfortable
thing in the world.

And they'd never
wear anything else.

But you can't find anybody
to make one of these things

now because they're
really labor intensive.

And the VA actually did
a whole bunch of research

in the '40s, trying
to figure out

how to make them not stink
because of they're basically

like shoes.

So we now have a pending
research proposal

that we should hear
from the Army and DARPA

from any day now to use current
athletic shoe technology to try

to make a flexible, variable
compliance socket that's

stiff in some places
and flexible and others,

breathable everywhere.

That basically solves
all the same problems

that your running shoes do.

And this would address
the number one reason

that people abandon
prosthetic arms.

And that, by the way, actually
includes, for example,

one of the DARPA arms that
was tested in VA clinics

among transradial
patients like me.

Comfort was actually
one of the major reasons

that I think about
half of those patients

said that they might not
be interested in having

one of those new arms.

So if we solve that
problem, then we

could potentially make a
lot of the other technology

more effective because
we'll be better

at attaching it to
people's bodies.

So that's what we're
up to right now.

An orphan device law.

So at the time that the orphan
drug law was passed in 1983,

we had a bipartisan Congress.

And I think that to imagine
in the current legislative

environment that we
can get anything done,

may be over optimistic.

Although, perhaps,
in an ideal world,

we would consider a law that
could increase innovation

and competition in this space.

Now, for those of you who
are unfamiliar with it, what

the orphan drug law does is, it
extends exclusivity for a drug

company who has
developed a product

but who believes that
the further investment

necessary to actually bring
a proven drug to market

is not worth the investment
because their patent,

the exclusivity available to
them on the original patent,

has expired.

Drug development is often
a really long process.

And it's a race against
the sundown on the patents.

Now unfortunately,
I'm not convinced

that a similar model in the this
even much smaller orphan device

spaces would
actually do anything

because exclusivity in
prosthetic arm patents

is often abandoned
when people decide

to stop paying the maintenance
on their process arm patents

after seven or 15 years.

So they're leaving exclusivity
on the table with regard

to patents in the space already.

So would extending it
actually do anything?

And I'm not convinced
that it would.

So Yeah?

AUDIENCE: [INAUDIBLE].

JON KUNIHOLM: OK.

So since we're not
miked in the audience,

the question is, is the approval
process for medical devices,

prosthetic arm
specifically, similarly long

as it is for drugs.

That's sort of a
complicated question

because it depends on
which class of devices

you're talking about.

The FDA it has three classes
of medical devices- I, II,

and III.

The DARPA neural sensing
system would be a Class III

medical device
because it actually

involves an internal implant.

Most prosthetic devices
right now are Class I.

And they're not only
Class I, they're

Class I Exempt which means that
there are exempt from-- there's

something like seven or eight
requirements for Class I

devices.

And all but three
of those are ignored

for this group of
Class I Exempt devices.

So in fact-- and this
is something actually

that there's been some--
some of the folks who've

been pursuing these new
things have been assuming--

and they may be right.
--that the FDA is seeking

to regulate prosthetic arm
stuff more rigorously than they

have in the past.

So even though you
could bring-- there

are articulated
myoelectric hands

that have been brought to
market in the last five years

that were brought to market.

All you have to do basically
is announced to the FDA

that you're going to start
selling these things and do it.

And if you're going to test
it, you stamp it experimental.

And you're good to go.

But nobody wants to invest
hundreds of millions of dollars

in the creation of
a device unless they

are sure that they're going
to get away with that.

And the FDA had
been making noises

that they were going
to try to bump up

the classification of
some of these things.

And they also have a requirement
for orthopedic devices

which is reasonable.

It's based on-- so a screw
in an orthopedic implant

is a Class I device that's a
component of a Class III device

which is the implant.

And there was a bad
failure in the '80s because

of the changing of
these Class I devices.

And so the FDA requires
each Class I component

be separately recertified
with the Class III system.

And if somebody did something
like that, for example,

with, say, there's a bolt
that can go through your skin

into the bone.

It's called osseointegration.

A titanium bolt.

And the bolt, of course,
is a Class III device

because it goes inside the body.

And if they required you
to separately certify

every prosthetic hand
or foot that you attach

to one of these systems,
it would effectively

mean that there wouldn't be
any available because nobody

would bother to do it.

So the whole-- so actually, you
bring up a really great point

which is that the FDA is a
really important part of this.

And some orphan device law, a
hypothetical orphan device law

or revisions of regulations that
might really impact the space,

should try to remove
some of those barriers

that actually the
FDA could potentially

create in trying to
better protect us.

And there are humanitarian
exemptions available.

And so it's something
that I think

bares further consideration.

So that leads into
the next group

of things which
are beyond new law.

What about the interpretation of
existing laws and regulations?

The Bayh-Dole Act
is the law from 1980

that governs the
government-funded creation

of intellectual property.

And the part that most of us are
familiar with, if you've ever

received a government grant,
is that the government

has royalty-free nonexclusive
license to anything

that they pay you to create.

But then, in order to
encourage economic activity,

anybody who creates this
stuff has the permission

to go off and license it.

Do whatever they want, exploit
that intellectual property

on their own, as long as they
don't charge any royalties back

to the government.

In practice, I think some
of this actually goes on.

Because if they fund,
particularly weapon systems

and stuff like that, it's
not like they segment off

what the intellectual
property component of it is.

It's not like anybody else can
make some of these fighter jets

or tanks or whatever.

So the least popular
provision of this law

is what are called
march-in rights.

Which means that if the
government-- if there are--

there are four criteria
that could be met where

the government can
come in and say,

you are not making use of this
intellectual property the way

we intended you to.

We're going to take it from you
and give it to somebody else.

And people-- it has been
exercised very few times,

if ever.

And it's a very unpopular,
as you can imagine,

provision with people who get
a lot of government funding.

How much time do we have?

So while this law
exists, it is something

that could be made use of
that's on the books already.

And so people, program
managers could say, OK,

so it's been 15 years.

You haven't-- you created
a product where we're going

to find somebody who will
because our goal is to bring

something like this to market.

And another one that
I referenced earlier

is the Center for Medicare
and Medicaid Services.

So all these devices, and many
medical devices are like this,

are reimbursed as durable
medical equipment-

canes, walkers, CPAP machines,
and prosthetic devices.

And the weirdness
that this creates

is that the whole
cost of supporting

the device for something
like five years

in between replacements
has to be captured just

in the provision of the product
at the beginning of its life

cycle.

All of the appointment
and everything.

So a prosthetist needs to be
able to function based on that.

The other thing that's
weird about this

is that there's a perverse
incentive, of course,

to provide more
of these products.

Because if you give
somebody a myoelectric arm,

it reimburses at
$30,000, $35,000.

A body-powered
arm may be $8,000.

And the other thing is that the
way that these reimbursement

codes are created is
that a company has

to demonstrate efficacy.

And then, they petition CMS.

And so that they have to
basically outlay all of the R&D

ahead before they know
if they're going to get

reimbursed.

Then, there have been
some recent cases

where people get the wrong
answer back from CMS.

And it kills the company.

A good example is
the iBot wheelchair

which everybody agrees is
pretty impressive technology.

That came back as being
excluded to a very

few pretty serious
disabilities and not

reimbursable at a rate that
could sustain its manufacturer.

And the company has
closed it doors.

And so it's actually
up in the air

what's going to happen
to that technology.

Hopefully, DEKA, which
is the creator of it,

will figure something out.

So one thing that I
think people might do

would be to create incentives
in reimbursement codes.

You could create a code
for the performance

of a particular device without
knowing that it exists.

And so I could say, a 15 degree
of freedom prosthetic hand

and wrist that's capable of
all these performance criteria

will reimburse at this rate,
$150,000, whatever it is.

And then, a company
would know ahead of time

that they have a
payday ahead of them

if they're capable of
meeting the criteria.

And so it would be much more
clear that this was possible.

Then, contract language.

There's a bunch of open source
and open architectural language

that the Navy, for one,
still has on kind of dead

website that's still up there.

But it's sample program
manager language

to put in RFPs to
ensure that you

can get what you
want out of them.

And those are tools
that could be used.

And then lastly, the DARPA
has made really good use

of challenges and contests.

And I think this sort of thing,
in partnership with industry,

might be very helpful.

So I guess in some
sense, my graphic here

of somebody cutting
red tape, maybe

I'm actually talking
about creating

some more in some cases.

But you get the picture.

So what about industry?

And that's you guys
here at Google.

I think that Google has done
some really creative things

in terms of business model in
creating great products that

solve problems in ways
that people hadn't before

by focusing the
business model in sort

of unexpected parts of it.

And so I think both Android,
both the Android operating

system and Gmail are
good examples of that.

In a recent acquisition of
yours, the nest company that

makes these thermostats
and the smoke detector,

those are examples
of products where

people would have said
the innovation is--

these are commodities.

You can't improve on this.

It is what it is.

Everybody has to have one.

And you're going to buy
the $8.95 one instead

of the $9.95 one because it just
does whatever it needs to do.

And I think the way that that
company was able to really pay

attention to consumers
represents creative thinking

which is the way that
probably everybody needs

to attack challenging markets.

And these ones that
I've talked about

are certainly some of them.

So you guys have
also been in the news

recently for, by count,
something like nine,

counting the drone company,
robotic acquisitions

over the last year or so.

And one of these
is Redwood Robotics

which I think involves a part
of a Willow Garage spin-off.

And I looked a little
bit-- I'd never

seen this hand picture before.

It looks to me like a rendering.

I don't know if it
represents a real product.

I think that it's
interesting that, in terms

of a manipulation and
terminal device, a hand

acquisition that you
guys chose to acquire,

Redwood and not any
of the companies

that were involved in the
DARPA prosthetic arms,

the-- there's an interesting
companionship between robot

teleoperation and prosthetics.

Myron Diftler, who runs the
robonaut program for NASA

at Johnson Space Center, said
that when he thought about it,

the prosthetic
problem is basically

a teleoperation problem.

You just happened to be
attached to the device you're

trying to teleoperate.

So in some sense, when you
take away video latency

to the moon or whatever, the
prosthetic control problem

might be, in some sense, easier.

Although the way that
these things are usually

performed right now is
with a haptic armature.

And in the prosthetic
case, obviously, you're

trying to control a limb that's
absent, so you don't have

access to those
fingers to tell you,

to tell those digits
where to move.

Anyway, my point
in bringing this up

is that there is some
great companionship

to these problems.

And in fact, while
it is not clear

right now who in America is
ready to buy, for example,

a personal assistance
robot or some other things

that one might
imagine as products

that could come out of companies
like these down the road.

There are prosthetic
customers who

can, according to the
current reimbursement

system, in some
cases, pay in excess

of $100,000 for one
of these devices.

So it might be possible to
create a synergy of interest

where you're solving problems
for the future of home

and mass-produced
robotics at the same time

as you are serving prosthetic
customers in the near term.

Even though they don't represent
the huge revenue stream

that we all would if everybody
has a Jetsons-style maid

in their home.

This is a picture
of my friend Kevin

who was born without
both of his legs.

I throw his picture up there
because he is, in some ways,

represents an orphan of an even
smaller community than my own.

Without any hip joints, when
he was about 10 years old,

they tried to put
prosthetic legs on him.

And they weren't really
anything but another couple

of extra crutches for him
to sort of haul around.

And he very quickly decided
that he wanted no part of that.

And he gets around
on a skateboard.

Now, that said, if
you were to envision--

and that's a way to
stay active, too.

But if you were to envision a
robotic prosthetic device that

might help him, it
could look something

like Boston Robotics,
Boston Dynamics-- I forget

the name of the
company. --in Cambridge.

They have the big
dog, walking bot

and those others where they
pretty famously kicked them

over in the video.

A pair of those walking legs,
controlled like the Segway,

is by leaning, could
do a pretty fair bit

towards replacing legs.

And so that's another case where
one of these robotics companies

might hold, in currently
available technology, the keys

to making a change
in somebody's life.

And my hope in
mentioning industry--

and I think, again, none of
these are mutually exclusive.

I do believe that
the only way that we

are going to solve these
extremely vexing problems

to society is by throwing nearly
everything that we have at it

and seeing what sticks.

And so really my purpose
in talking to you today

is to try-- I want to start
this conversation because it's

pretty clear to me that there
are some things that we could

do better about how we're
responding to these problems.

And I'd very much like some help
in figuring out how to do it.

So I used up more time
than I had intended to.

But if you all
have any questions,

I'd certainly like to
try and answer them.

So the question is-- I mentioned
that most arm amputees still

use hooks vise the
more complicated

electrical externally-powered
prostheses and why.

I think cost is part of it.

The Affordable Care Act
has changed some of that.

Before January, a lot of
people with medical insurance

had lifetime caps of, in
some cases, just more,

like $1,500 lifetime cap on
prosthetics services which

isn't even enough to
get a single prosthesis.

And beyond the
expansion of coverage,

the removal of the lifetime
caps and pre-existing conditions

are huge for the
prosthetic community.

So to the extent
that it is monetary,

the Affordable Care Act
should change some of that.

I, anecdotally, I don't
think that there's

any great data on this.

But I think that a lot
of common complaints

about the myoelectric
prostheses are latencies,

unintended movements, the speed
with which the motors move,

the weight.

And then, I think you
hit on it, function

is probably the primary reason.

The hook has been famously
described in TED Talks

as a rubber band and a
hook on the end of a stick.

But I believe that
the bar is actually

quite a bit higher than that.

The Dorrance hook is really
kind of a marvel of design.

It's got a bunch of
affordances built into it.

You can pick something
up, use it as a hook,

and as a split hook,
you can open it up

and use the individual
fingers for different things.

The little thumb that
has the cable on it---

I'll put that back up.

The little thumb that
has the cable on it

can be used to push things.

And is the part of
the device that's

used to wedge a
knife and a fork.

Actually that's
a perfect example

is eating with a knife and fork.

With cosmetic coverings
on the myoelectric hands,

they tend to be slippery.

And it's hard to get a knife and
a fork to sit securely in them.

And this thing is designed
to do exactly that.

It's got a chisel
tang in the middle,

so you can stick
a tool in there.

Or you can wedge either
a knife or a fork

in between the fingers
and over the thumb.

So I think function
is a lot of it.

Any other questions?

Yeah?

AUDIENCE: [INAUDIBLE].

JON KUNIHOLM: Yeah.

So well, first of all,
it's selfish of me, right?

That's my problem.

So that's what I spend a
lot of time thinking about.

But there are other
complications.

So there are something like
50 times more leg amputees

in America than arm amputees
because of peripheral vascular

disease secondary diabetes.

And that's the major
cause of leg amputation.

So there are more than
two million leg amputees

in America.

And most of-- and is an
easier problem to solve, too.

When you're looking at--
well, Oscar Pistorius

is a great example.

The cheetah legs make somebody
capable of running close to 10

second 100.

Although knees are a problem.

And Oscar Pistorius is
a below knee amputee.

But then, among the
powered devices,

control engineers
who work on legs

have told me that it's possible
to do reasonable approximations

of gate with the individual
components not even talking

to each other.

Where they just respond
to the external forces,

and they're capable of doing
a pretty good job walking.

And contrast that to
an arm where just,

forget the manipulation
stuff, just

reach to grasp with a
whole arm for somebody

up to the shoulder level
involves the coordination

of all of the joints together.

So I like to say that if you
imagine that these are legs,

the arm problem has five of
them on the end of a sixth.

And they all have to
dance together in order

to do the most basic task
that an arm has to do.

Yeah.

And so the economics are--
legs are a lot bigger business.

So there is a real incentive.

Although there are
some new leg companies,

because of the high cost of
additional degrees of freedom,

there are some
leg companies that

are going to be in
trouble if they don't

get the reimbursements that
they want on powered ankles,

for example.

And so in that sense they
share exactly the same problem.

Anything else?

Any questions from our
remote rooms?

All right.

Well, listen.

Thank you very much for taking
the time to check it out.

And I don't if there's a way for
me to post contact information

or whatever.

But if anybody's interested
in further talking about this,

please get with
Michael Weiss-Malik.

And he'll put you
in touch with me.

Thanks very much.

(Applause)