WEBVTT

00:00:00.591 --> 00:00:03.096
Five years ago, I was a Ph.D student

00:00:03.096 --> 00:00:04.880
living two lives.

00:00:04.880 --> 00:00:07.396
In one, I used NASA supercomputers

00:00:07.396 --> 00:00:09.985
to design next generation spacecraft,

00:00:09.985 --> 00:00:12.381
and in the other I was a data scientist

00:00:12.381 --> 00:00:14.937
looking for potential smugglers

00:00:14.937 --> 00:00:17.977
of sensitive nuclear technologies.

00:00:17.977 --> 00:00:21.191
As a data scientist, I did a lot of analyses,

00:00:21.191 --> 00:00:22.688
mostly of facilities,

00:00:22.688 --> 00:00:24.696
industrial facilities around the world.

00:00:24.696 --> 00:00:27.571
And I was always looking for a better canvas

00:00:27.571 --> 00:00:29.512
to tie these all together.

00:00:29.512 --> 00:00:31.501
And one day, I was thinking about how

00:00:31.501 --> 00:00:33.563
all data has a location,

00:00:33.563 --> 00:00:35.128
and I realized that the answer

00:00:35.128 --> 00:00:37.108
had been staring me in the face.

00:00:37.108 --> 00:00:39.776
Although I was a satellite engineer,

00:00:39.776 --> 00:00:42.824
I hadn't thought about using satellite imagery

00:00:42.824 --> 00:00:44.513
in my work.

NOTE Paragraph

00:00:44.513 --> 00:00:46.399
Now like most of us, I'd been online,

00:00:46.399 --> 00:00:48.273
I'd see my house, so I thought,

00:00:48.273 --> 00:00:50.296
I'll hop in there and I'll start looking up

00:00:50.296 --> 00:00:51.958
some of these facilities.

00:00:51.958 --> 00:00:53.880
And what I found really surprised me.

00:00:53.880 --> 00:00:55.616
The pictures that I was finding

00:00:55.616 --> 00:00:57.501
were years out of date,

00:00:57.501 --> 00:00:58.585
and because of that,

00:00:58.585 --> 00:01:00.194
it had relatively little relevance

00:01:00.194 --> 00:01:02.731
to the work that I was doing today.

00:01:02.731 --> 00:01:04.192
But I was intrigued.

00:01:04.192 --> 00:01:07.428
I mean, satellite imagery is pretty amazing stuff.

00:01:07.428 --> 00:01:09.731
There are millions and millions of sensors

00:01:09.731 --> 00:01:11.129
surrounding us today,

00:01:11.129 --> 00:01:14.320
but there's still so much we
don't know on a daily basis.

00:01:14.320 --> 00:01:17.689
How much oil is stored in all of China?

00:01:17.689 --> 00:01:20.529
How much corn is being produced?

00:01:20.529 --> 00:01:24.815
How many ships are in all of our world's ports?

00:01:24.815 --> 00:01:27.390
Now, in theory, all of these questions

00:01:27.390 --> 00:01:29.311
could be answered by imagery,

00:01:29.311 --> 00:01:31.391
but not if it's old.

00:01:31.391 --> 00:01:33.895
And if this data was so valuable,

00:01:33.895 --> 00:01:35.751
then how come I couldn't get my hands

00:01:35.751 --> 00:01:38.377
on more recent pictures?

NOTE Paragraph

00:01:38.377 --> 00:01:41.543
So the story begins over 50 years ago

00:01:41.543 --> 00:01:43.399
with the launch of the first generation

00:01:43.399 --> 00:01:46.783
of U.S. government photo reconnaissance satellites.

00:01:46.783 --> 00:01:48.495
And today, there's a handful

00:01:48.495 --> 00:01:50.567
of the great, great grandchildren

00:01:50.567 --> 00:01:52.416
of these early cold war machines

00:01:52.416 --> 00:01:54.472
which are now operated by private companies

00:01:54.472 --> 00:01:57.119
and from which the vast majority of satellite imagery

00:01:57.119 --> 00:01:59.975
that you and I see on a daily basis comes.

00:01:59.975 --> 00:02:03.072
During this period, launching things into space,

00:02:03.072 --> 00:02:05.736
just the rocket to get the satellite up there,

00:02:05.736 --> 00:02:10.029
has cost hundreds of millions of dollars each,

00:02:10.029 --> 00:02:11.889
and that's created tremendous pressure

00:02:11.889 --> 00:02:14.377
to launch things infrequently

00:02:14.377 --> 00:02:15.706
and to make sure that when you do,

00:02:15.706 --> 00:02:19.036
you cram as much functionality in there as possible.

00:02:19.036 --> 00:02:20.802
All of this has only made satellites

00:02:20.802 --> 00:02:23.028
bigger and bigger and bigger

00:02:23.028 --> 00:02:24.786
and more expensive,

00:02:24.786 --> 00:02:30.050
now nearly a billion, with a b, dollars per copy.

00:02:30.050 --> 00:02:31.722
Because they are so expensive,

00:02:31.722 --> 00:02:33.042
there aren't very many of them.

00:02:33.042 --> 00:02:34.426
Because there aren't very many of them,

00:02:34.426 --> 00:02:36.775
the pictures that we see on a daily basis

00:02:36.775 --> 00:02:38.386
tend to be old.

NOTE Paragraph

00:02:38.386 --> 00:02:41.514
I think a lot of people actually
understand this anecdotally,

00:02:41.514 --> 00:02:43.961
but in order to visualize just how sparsely

00:02:43.961 --> 00:02:45.738
our planet is collected,

00:02:45.738 --> 00:02:48.250
some friends and I put together a data set

00:02:48.250 --> 00:02:50.530
of the 30 million pictures that have been gathered

00:02:50.530 --> 00:02:54.259
by these satellites between 2000 and 2010.

00:02:54.259 --> 00:02:56.829
As you can see in blue, huge areas of our world

00:02:56.829 --> 00:02:59.619
are barely seen, less than once a year,

00:02:59.619 --> 00:03:01.968
and even the areas that are seen most frequently,

00:03:01.968 --> 00:03:05.639
those in red, are seen at best once a quarter.

00:03:05.639 --> 00:03:08.679
Now as aerospace engineering grad students,

00:03:08.679 --> 00:03:12.214
this chart cried out to us as a challenge.

00:03:12.214 --> 00:03:15.715
Why do these things have to be so expensive?

00:03:15.715 --> 00:03:18.297
Does a single satellite really have to cost

00:03:18.297 --> 00:03:22.936
the equivalent of three 747 jumbo jets?

00:03:22.936 --> 00:03:25.238
Wasn't there a way to build a smaller,

00:03:25.238 --> 00:03:28.271
simpler, new satellite design that could enable

00:03:28.271 --> 00:03:30.382
more timely imaging?

NOTE Paragraph

00:03:30.382 --> 00:03:34.031
Now I realize that it does sound a little bit crazy

00:03:34.031 --> 00:03:35.514
that we were going to go out and just

00:03:35.514 --> 00:03:37.191
begin designing satellites,

00:03:37.191 --> 00:03:39.200
but fortunately we had help.

00:03:39.200 --> 00:03:42.104
In the late 1990s, a couple of professors

00:03:42.104 --> 00:03:45.616
proposed a concept for radically reducing the price

00:03:45.616 --> 00:03:47.311
of putting things in space.

00:03:47.311 --> 00:03:49.423
This was hitchhiking small satellites

00:03:49.423 --> 00:03:52.392
alongside much larger satellite.

00:03:52.392 --> 00:03:55.563
This dropped the cost of putting objects up there

00:03:55.563 --> 00:03:57.407
by over a factor of 100,

00:03:57.407 --> 00:04:00.585
and suddenly we could afford to experiment,

00:04:00.585 --> 00:04:01.991
to take a little bit of risk,

00:04:01.991 --> 00:04:04.303
and to realize a lot of innovation.

00:04:04.303 --> 00:04:07.423
And a new generation of engineers and scientists,

00:04:07.423 --> 00:04:08.991
mostly out of universities,

00:04:08.991 --> 00:04:11.320
began launching these very small,

00:04:11.320 --> 00:04:13.881
breadbox-sized satellites called CubeSats.

00:04:13.881 --> 00:04:16.056
And these were built with electronics obtained

00:04:16.056 --> 00:04:20.119
from RadioShack instead of Lockheed Martin.

NOTE Paragraph

00:04:20.119 --> 00:04:22.836
Now it was using the lessons
learned from these early missions

00:04:22.836 --> 00:04:25.516
that my friends and I began a series of sketches

00:04:25.516 --> 00:04:27.420
of our own satellite design.

00:04:27.420 --> 00:04:30.062
And, you know, I can't remember a specific day

00:04:30.062 --> 00:04:31.972
where we made, like, a conscious decision

00:04:31.972 --> 00:04:34.653
that we were actually going to
go out and build these things,

00:04:34.653 --> 00:04:37.099
but it was just, once we got that idea in our minds

00:04:37.099 --> 00:04:39.012
of the world as a data set,

00:04:39.012 --> 00:04:41.610
of being able to capture millions of data points

00:04:41.610 --> 00:04:44.387
on a daily basis, describing the global economy,

00:04:44.387 --> 00:04:46.892
of being able to unearth billions of connections

00:04:46.892 --> 00:04:50.196
between them that had never before been found,

00:04:50.196 --> 00:04:52.196
it just seemed boring

00:04:52.196 --> 00:04:54.612
to go work on anything else.

NOTE Paragraph

00:04:54.612 --> 00:04:57.892
And so we moved into a cramped,

00:04:57.892 --> 00:05:01.061
windowless office in Palo Alto,

00:05:01.061 --> 00:05:02.780
and began working to take our design

00:05:02.780 --> 00:05:05.976
from the drawing board into the lab.

00:05:05.976 --> 00:05:08.572
Now the first major question we had to tackle

00:05:08.572 --> 00:05:10.860
was just how big to build this thing.

00:05:10.860 --> 00:05:14.012
In space, size drives cost,

00:05:14.012 --> 00:05:16.078
and we had worked with these very small,

00:05:16.078 --> 00:05:18.228
breadbox-sized satellites in school,

00:05:18.228 --> 00:05:20.732
but as we began to better
understand the laws of physics,

00:05:20.732 --> 00:05:22.715
we found that the quality of pictures

00:05:22.715 --> 00:05:25.707
those satellites could take was very limited,

00:05:25.707 --> 00:05:27.596
because the laws of physics dictate

00:05:27.596 --> 00:05:30.748
that the best picture you
can take through a telescope

00:05:30.748 --> 00:05:33.068
is a function of the diameter of that telescope,

00:05:33.068 --> 00:05:34.572
and these satellites had a very small,

00:05:34.572 --> 00:05:36.601
very constrained volume.

00:05:36.601 --> 00:05:38.227
And we found that the best picture we would

00:05:38.227 --> 00:05:40.438
have been able to get looked something like this.

00:05:40.438 --> 00:05:42.557
And although this was the low-cost option,

00:05:42.557 --> 00:05:44.461
quite frankly it was just too blurry

00:05:44.461 --> 00:05:48.277
to see the things that make
satellite imagery valuable.

00:05:48.277 --> 00:05:50.253
So about three or four weeks later,

00:05:50.253 --> 00:05:53.101
we met a group of engineers randomly

00:05:53.101 --> 00:05:54.757
who had worked on the first

00:05:54.757 --> 00:05:57.136
private imaging satellite ever developed,

00:05:57.136 --> 00:05:59.301
and they told us that back in the 1970s,

00:05:59.301 --> 00:06:01.099
the U.S. government had found a powerful

00:06:01.099 --> 00:06:03.125
optimal tradeoff,

00:06:03.125 --> 00:06:06.229
that in taking pictures at right
about one meter resolution,

00:06:06.229 --> 00:06:08.944
being able to see objects one meter in size,

00:06:08.944 --> 00:06:11.773
they found that they could not
just get very high quality images,

00:06:11.773 --> 00:06:14.681
but get a lot of them at an affordable price.

00:06:14.681 --> 00:06:16.473
From our own computer simulations,

00:06:16.473 --> 00:06:18.693
we quickly found that one meter really was

00:06:18.693 --> 00:06:20.365
the minimum viable product

00:06:20.365 --> 00:06:23.428
to be able to see the drivers of our global economy,

00:06:23.428 --> 00:06:24.974
for the first time, being able to count

00:06:24.974 --> 00:06:27.557
the ships and cars and shipping
containers and trucks

00:06:27.557 --> 00:06:30.550
that move around our world on a daily basis,

00:06:30.550 --> 00:06:34.477
while conveniently still not
being able to see individuals.

00:06:34.477 --> 00:06:36.221
We had found our compromise.

00:06:36.221 --> 00:06:37.518
We would have to build something larger

00:06:37.518 --> 00:06:39.189
than the original breadbox,

00:06:39.189 --> 00:06:40.809
now more like a mini-fridge,

00:06:40.809 --> 00:06:43.554
but we still wouldn't have to build a pickup truck.

00:06:43.554 --> 00:06:45.642
So now we had our constraint.

00:06:45.642 --> 00:06:47.566
The laws of physics dictated

00:06:47.566 --> 00:06:51.069
the absolute minimum-sized
telescope that we could build.

NOTE Paragraph

00:06:51.069 --> 00:06:54.009
What came next was making the rest of the satellite

00:06:54.009 --> 00:06:56.101
as small and as simple as possible,

00:06:56.101 --> 00:06:58.613
basically a flying telescope with four walls

00:06:58.613 --> 00:07:01.693
and a set of electronics smaller than a phone book

00:07:01.693 --> 00:07:04.992
that used less power than a 100 watt light bulb.

00:07:04.992 --> 00:07:06.646
The big challenge became actually taking

00:07:06.646 --> 00:07:09.214
the pictures through that telescope.

00:07:09.214 --> 00:07:12.134
Traditional imaging satellites use a line scanner,

00:07:12.134 --> 00:07:13.895
similar to a Xerox machine,

00:07:13.895 --> 00:07:16.295
and as they traverse the earth, they take pictures,

00:07:16.295 --> 00:07:18.615
scanning row by row by row

00:07:18.615 --> 00:07:20.398
to build the complete image.

00:07:20.398 --> 00:07:23.447
Now people use these because they get a lot of light,

00:07:23.447 --> 00:07:25.217
which means less of the noise you see

00:07:25.217 --> 00:07:27.814
in a low-cost cell phone image.

00:07:27.814 --> 00:07:29.575
The problem with them is they require

00:07:29.575 --> 00:07:32.446
very sophisticated pointing.

00:07:32.446 --> 00:07:34.899
You have to stay focused on a 50-centimeter target

00:07:34.899 --> 00:07:36.683
from over 600 miles away

00:07:36.683 --> 00:07:39.030
while moving at more than
seven kilometers a second,

00:07:39.030 --> 00:07:41.510
which requires an awesome degree of complexity.

00:07:41.510 --> 00:07:45.193
So instead, we turned to a new
generation of video sensors,

00:07:45.193 --> 00:07:48.160
originally created for use in night vision goggles.

00:07:48.160 --> 00:07:51.029
Instead of taking a single, high quality image,

00:07:51.029 --> 00:07:52.446
we could take a videostream

00:07:52.446 --> 00:07:55.631
of individually noisier frames,

00:07:55.631 --> 00:07:57.267
but then we could recombine

00:07:57.267 --> 00:07:58.795
all of those frames together

00:07:58.795 --> 00:08:00.699
into very high quality images

00:08:00.699 --> 00:08:03.476
using sophisticated pixel processing techniques

00:08:03.476 --> 00:08:04.843
here on the ground,

00:08:04.843 --> 00:08:08.119
at a cost of one one-hundredth a traditional system.

00:08:08.119 --> 00:08:09.287
And we applied this technique

00:08:09.287 --> 00:08:12.080
to many of the other systems on the satellite as well,

00:08:12.080 --> 00:08:15.132
and day by day, our design evolved

00:08:15.132 --> 00:08:18.282
from cad to prototypes

00:08:18.282 --> 00:08:21.140
to production units.

NOTE Paragraph

00:08:21.140 --> 00:08:22.827
A few short weeks ago,

00:08:22.827 --> 00:08:25.155
we packed up SkySat 1,

00:08:25.155 --> 00:08:26.651
put our signatures on it,

00:08:26.651 --> 00:08:28.859
and waved goodbye for the last time on earth.

00:08:28.859 --> 00:08:32.155
Today, it's sitting in its final launch configuration

00:08:32.155 --> 00:08:35.509
ready to blast off in a few short weeks.

00:08:35.509 --> 00:08:37.682
And soon, we'll turn our attention to launching

00:08:37.682 --> 00:08:40.851
a constellation of 24 or more of these satellites

00:08:40.851 --> 00:08:43.139
and beginning to build the scalable analytics

00:08:43.139 --> 00:08:45.509
that will allow us to unearth the insights

00:08:45.509 --> 00:08:48.667
in the pedabytes of data we will collect.

NOTE Paragraph

00:08:48.667 --> 00:08:52.779
So why do all of this? Why build these satellites?

00:08:52.779 --> 00:08:55.251
Well, it turns out imaging satellites

00:08:55.251 --> 00:08:59.100
have a unique ability to provide global transparency,

00:08:59.100 --> 00:09:01.634
and providing that transparency on a timely basis

00:09:01.634 --> 00:09:04.922
is simply an idea whose time has come.

00:09:04.922 --> 00:09:08.338
We see ourselves as pioneers of a new frontier,

00:09:08.338 --> 00:09:10.547
and beyond economic data,

00:09:10.547 --> 00:09:14.365
unlocking the human story, moment by moment.

00:09:14.365 --> 00:09:15.619
For a data scientist

00:09:15.619 --> 00:09:18.230
that just happened to go to space camp as a kid,

00:09:18.230 --> 00:09:21.130
it just doesn't get much better than that.

NOTE Paragraph

00:09:21.130 --> 00:09:23.283
Thank you.

NOTE Paragraph

00:09:23.283 --> 00:09:27.263
(Applause)