WEBVTT

00:00:00.891 --> 00:00:02.854
Every single one of us will lose

00:00:02.854 --> 00:00:06.011
or has already lost something
we rely on every single day.

00:00:06.721 --> 00:00:09.222
I am of course talking about our keys.

00:00:09.530 --> 00:00:10.931
(Laughter)

00:00:11.194 --> 00:00:12.210
Just kidding.

00:00:12.210 --> 00:00:15.542
What I actually want to talk about
is one of our most important senses:

00:00:15.542 --> 00:00:16.536
vision.

00:00:16.536 --> 00:00:19.260
Every single day we each lose
a little bit of our ability

00:00:19.260 --> 00:00:20.594
to refocus our eyes

00:00:20.594 --> 00:00:22.328
until we can't refocus at all.

00:00:22.837 --> 00:00:24.960
We call this condition presbyopia,

00:00:24.960 --> 00:00:27.324
and it affects two billion
people worldwide.

00:00:27.554 --> 00:00:29.324
That's right, I said billion.

00:00:29.524 --> 00:00:31.204
If you haven't heard of presbyopia,

00:00:31.204 --> 00:00:34.010
and you're wondering, 
"Where are these two billion people?"

00:00:34.010 --> 00:00:36.074
here's a hint before
I get into the details.

00:00:36.074 --> 00:00:39.217
It's the reason why people wear
reading glasses or bifocal lenses.

00:00:39.537 --> 00:00:42.435
I'll get started by describing
the loss in refocusing ability

00:00:42.435 --> 00:00:43.774
leading up to presbyopia.

00:00:43.952 --> 00:00:44.968
As a newborn,

00:00:44.968 --> 00:00:48.415
you would have been able to focus
as lose as six-and-a-half centimenters

00:00:48.415 --> 00:00:49.636
if you wish to.

00:00:49.636 --> 00:00:52.698
By your mid-20s, you have about
half of that focusing power left.

00:00:52.734 --> 00:00:53.950
10 centimeters or so,

00:00:53.950 --> 00:00:56.485
but close enough that you never
notice the difference.

00:00:56.485 --> 00:00:57.640
By your late-40s though,

00:00:57.640 --> 00:01:00.008
the closest you can focus
is about 25 centimeters,

00:01:00.008 --> 00:01:01.242
maybe even farther.

00:01:01.242 --> 00:01:03.316
Losses in focusing ability
beyond this point

00:01:03.316 --> 00:01:05.526
start effecting near-vision
tasks like reading,

00:01:05.526 --> 00:01:07.489
and by the time you reach age 60,

00:01:07.489 --> 00:01:09.992
nothing within a meter
radius of you is clear.

00:01:10.168 --> 00:01:12.335
Right now some of you
are probably thinking,

00:01:12.335 --> 00:01:15.683
that sounds bad but he means
you in a figurative sense,

00:01:15.683 --> 00:01:18.622
only for the people that actually
end up with presbyopia.

00:01:19.128 --> 00:01:23.578
But no, when I say you, I literally mean
that every single one of you

00:01:23.578 --> 00:01:26.429
will someday be presbyopic
if you aren't already.

00:01:26.852 --> 00:01:28.237
That sounds a bit troubling.

00:01:28.237 --> 00:01:31.901
I want to remind you that presbyopia
has been with us for all of human history

00:01:31.901 --> 00:01:34.724
and we've done a lot of different
things to try and fix it.

00:01:34.730 --> 00:01:38.457
So to start, let's imagine
that you're sitting at a desk, reading.

00:01:38.699 --> 00:01:40.091
If you were presbyopic,

00:01:40.091 --> 00:01:42.098
it might look a little
something like this.

00:01:42.098 --> 00:01:44.840
Anything close by,
like the magazine, will be blurry.

00:01:45.196 --> 00:01:46.454
Moving on to solutions.

00:01:46.454 --> 00:01:48.044
First, reading glasses.

00:01:48.285 --> 00:01:50.317
These have lenses
with a single focal power

00:01:50.328 --> 00:01:52.408
tuned so that near objects
come into focus.

00:01:52.609 --> 00:01:55.322
But far objects necessarily
go out of focus,

00:01:55.322 --> 00:01:57.763
meaning you have to constantly
switch back and forth

00:01:57.763 --> 00:01:59.535
between wearing and not wearing them.

00:01:59.535 --> 00:02:00.759
To solve this problem

00:02:00.759 --> 00:02:04.026
Benjamin Franklin invented
what he called "double spectacles."

00:02:04.241 --> 00:02:06.249
Today we call those bifocals,

00:02:06.249 --> 00:02:09.837
and what they let him do
was see far when he looked up

00:02:09.837 --> 00:02:11.459
and see near when he looked down.

00:02:11.459 --> 00:02:14.463
Today we also have progressive lenses
which get rid of the line

00:02:14.463 --> 00:02:17.033
by smoothly varying the focal power
from top to bottom.

00:02:17.033 --> 00:02:18.445
The downside to both of these

00:02:18.445 --> 00:02:21.244
is that you lose field of vision
at any given distance

00:02:21.244 --> 00:02:23.811
because it gets split up
from top to bottom like this.

00:02:23.811 --> 00:02:25.180
To see why that's a problem,

00:02:25.180 --> 00:02:27.975
imagine that you're climbing
down a ladder or stairs.

00:02:28.166 --> 00:02:31.325
You look down to get
your footing but it's blurry.

00:02:31.691 --> 00:02:33.083
Why would it be blurry?

00:02:33.297 --> 00:02:36.567
Well, you look down
and that's the near part of the lens,

00:02:36.567 --> 00:02:39.275
but the next step was past arm's reach,

00:02:39.275 --> 00:02:41.104
which for your eyes, counts as far.

00:02:41.274 --> 00:02:44.143
The next solution I want to point out
is a little less common

00:02:44.143 --> 00:02:46.757
but comes up a lot in contact lenses
or LASIK surgeries,

00:02:46.757 --> 00:02:48.070
and it's called monovision.

00:02:48.070 --> 00:02:50.590
It works by setting up
the dominant eye to focus far

00:02:50.590 --> 00:02:52.114
and the other eye to focus near.

00:02:52.114 --> 00:02:54.830
Your brain does the work
of intelligently putting together

00:02:54.830 --> 00:02:56.816
the sharpest parts from each eye's view,

00:02:56.816 --> 00:02:59.004
but the two eyes see
slightly different things,

00:02:59.004 --> 00:03:01.620
and that makes it harder
to judge distances binocularly.

00:03:01.620 --> 00:03:03.150
So where does that all leave us?

00:03:03.150 --> 00:03:06.001
It seems like we've come up
with a lot of different solutions

00:03:06.001 --> 00:03:08.338
but none of them quite restore
natural refocusing.

00:03:08.338 --> 00:03:10.334
None of them let you
just look at something

00:03:10.334 --> 00:03:11.738
and expect it to be in focus.

00:03:11.738 --> 00:03:12.736
But why?

00:03:12.840 --> 00:03:13.984
Well, to explain that

00:03:13.984 --> 00:03:16.693
we'll want to take a look
at the anatomy of the human eye.

00:03:16.693 --> 00:03:19.877
The part of the eye that allows us
to refocus to different distances

00:03:19.877 --> 00:03:21.398
is called the crystalline lens.

00:03:21.654 --> 00:03:25.644
There are muscles surrounding the lens
that can deform it into different shapes,

00:03:25.644 --> 00:03:27.794
which in turn changes its focusing power.

00:03:27.794 --> 00:03:29.913
What happens when someone
becomes presbyopic?

00:03:30.213 --> 00:03:32.399
It turns out that
the crystalline lens stiffens

00:03:32.399 --> 00:03:35.078
to the point that it doesn't
really change shape anymore.

00:03:35.078 --> 00:03:39.080
Now, thinking back
on all the solutions I listed earlier,

00:03:39.080 --> 00:03:42.686
we can see that they all have
something in common with the others

00:03:42.686 --> 00:03:44.311
but not with our eyes,

00:03:44.311 --> 00:03:46.204
and that is that they're all static.

00:03:46.290 --> 00:03:49.094
It's like the optical equivalent
of a pirate with a peg leg.

00:03:49.094 --> 00:03:51.990
What is the optical equivalent
of a modern prosthetic leg?

00:03:52.291 --> 00:03:55.524
The last several decades have seen
the creation and rapid development

00:03:55.524 --> 00:03:57.927
of what are called "focus-tunable lenses."

00:03:58.333 --> 00:04:00.020
There are several different types.

00:04:00.020 --> 00:04:01.813
Mechanically-shifted Alvarez lenses,

00:04:01.813 --> 00:04:03.447
deformable liquid lenses

00:04:03.447 --> 00:04:05.875
and electronically-switched, 
liquid crystal lenses.

00:04:05.959 --> 00:04:07.686
Now these have their own trade-offs,

00:04:07.686 --> 00:04:10.287
but what they don't skimp on
is the visual experience.

00:04:10.287 --> 00:04:13.618
Full, field-of-view vision that can be
sharp at any desirable distance.

00:04:13.618 --> 00:04:14.619
OK, great.

00:04:14.619 --> 00:04:16.193
The lenses we need already exist.

00:04:16.193 --> 00:04:17.293
Problem solved, right?

00:04:17.786 --> 00:04:19.002
Not so fast.

00:04:19.156 --> 00:04:22.054
Focus-tunable lenses add a bit
of complexity to the equation.

00:04:22.162 --> 00:04:25.981
The lenses don't have any way of knowing
what distance they should be focused to.

00:04:25.981 --> 00:04:28.887
What we really need are glasses
that, when you're looking far,

00:04:28.887 --> 00:04:30.001
far objects are sharp,

00:04:30.001 --> 00:04:31.153
and when you look near,

00:04:31.153 --> 00:04:33.924
near objects come into focus
anywhere in your field of view

00:04:33.924 --> 00:04:35.908
and without you having to think about it.

00:04:35.908 --> 00:04:38.375
What I've worked on
these last few years at Stanford

00:04:38.375 --> 00:04:40.914
is building that exact intelligence
around the lenses.

00:04:40.914 --> 00:04:44.480
Our prototype borrows technology
from virtual and augmented reality systems

00:04:44.480 --> 00:04:45.917
to estimate focusing distance.

00:04:45.917 --> 00:04:49.487
We have an eye tracker that can tell
what direction our eyes are focused in.

00:04:49.487 --> 00:04:50.488
Using two of these,

00:04:50.488 --> 00:04:52.691
we can directly triangulate
your gaze direction

00:04:52.691 --> 00:04:53.864
to get a focus estimate.

00:04:53.864 --> 00:04:54.864
Just in case though,

00:04:54.864 --> 00:04:57.483
to increase reliability
we also added a distance sensor.

00:04:57.483 --> 00:04:59.820
The sensor is a camera
that looks out at the world

00:04:59.820 --> 00:05:01.420
and reports distances to objects.

00:05:01.420 --> 00:05:04.420
We can again use your gaze direction
to get a distance estimate

00:05:04.420 --> 00:05:05.599
for a second time.

00:05:05.599 --> 00:05:07.580
We then fuse those two distance estimates

00:05:07.580 --> 00:05:10.075
and update the focus-tunable
lens power accordingly.

00:05:10.266 --> 00:05:13.592
The next step for us was of course
to test our device on actual people.

00:05:13.592 --> 00:05:16.680
So we recruited about 100 presbyopes
and had them test our device

00:05:16.680 --> 00:05:18.392
while we measured their performance.

00:05:18.392 --> 00:05:21.719
What we saw convinced us right then
that autofocals were the future.

00:05:21.719 --> 00:05:23.634
Our participants could see more clearly,

00:05:23.634 --> 00:05:25.023
they could focus more quickly

00:05:25.023 --> 00:05:28.031
and they thought it was an easier
and better focusing experience

00:05:28.031 --> 00:05:29.461
than their current correction.

00:05:29.461 --> 00:05:30.457
To put it simply,

00:05:30.457 --> 00:05:31.603
when it comes to vision,

00:05:31.603 --> 00:05:34.811
autofocals don't compromise
like static corrections in use today do.

00:05:34.811 --> 00:05:36.758
But I don't want to get ahead of myself.

00:05:36.758 --> 00:05:39.500
There's a lot of work
for my colleagues and me left to do.

00:05:39.500 --> 00:05:41.948
For example, our glasses are a bit --

00:05:41.948 --> 00:05:42.947
(Laughter)

00:05:42.947 --> 00:05:44.462
bulky, maybe?

00:05:44.462 --> 00:05:47.698
And one reason for this
is that we used bulkier components

00:05:47.698 --> 00:05:50.492
that are often intended
for research use or industrial use.

00:05:50.492 --> 00:05:52.754
Another is that we need
to strap everything down

00:05:52.754 --> 00:05:56.485
because current eye-tracking algorithms
don't have the robustness that we need.

00:05:56.776 --> 00:05:58.051
So moving forward,

00:05:58.051 --> 00:06:00.556
as we move from a research
setting into a start-up,

00:06:00.556 --> 00:06:02.618
we plan to make future autofocals

00:06:02.618 --> 00:06:05.171
eventually look a little bit more
like normal glasses.

00:06:05.171 --> 00:06:06.168
For this to happen,

00:06:06.168 --> 00:06:08.554
we'll need to significantly improve

00:06:08.554 --> 00:06:10.667
the robustness
of our eye-tracking solution.

00:06:10.667 --> 00:06:14.539
We'll also need to incorporate smaller
and more efficient electronics and lenses.

00:06:14.919 --> 00:06:17.211
That said, even with
our current prototype,

00:06:17.211 --> 00:06:19.762
we've shown that today's
focus-tunable lens technology

00:06:19.762 --> 00:06:23.345
is capable of out-performing
traditional forms of static correction.

00:06:23.345 --> 00:06:24.894
So it's only a matter of time.

00:06:25.134 --> 00:06:27.158
It's pretty clear that in the near future,

00:06:27.164 --> 00:06:30.168
instead of worrying about which pair
of glasses to use and when,

00:06:30.168 --> 00:06:32.725
we'll be able to just focus
on the important things.

00:06:33.687 --> 00:06:34.686
Thank you.

00:06:34.686 --> 00:06:36.438
(Applause)