0:00:00.506,0:00:01.913
So, how many of you have ever

0:00:01.913,0:00:03.568
gotten behind the wheel of a car

0:00:03.568,0:00:09.255
when you really shouldn't have been driving?

0:00:09.255,0:00:11.160
Maybe you're out on the road for a long day,

0:00:11.160,0:00:12.650
and you just wanted to get home.

0:00:12.650,0:00:15.297
You were tired, but you felt you could drive a few more miles.

0:00:15.297,0:00:16.496
Maybe you thought,

0:00:16.496,0:00:18.513
I've had less to drink than everybody else,

0:00:18.513,0:00:20.249
I should be the one to go home.

0:00:20.249,0:00:24.840
Or maybe your mind was just entirely elsewhere.

0:00:24.840,0:00:26.294
Does this sound familiar to you?

0:00:26.294,0:00:29.192
Now, in those situations, wouldn't it be great

0:00:29.192,0:00:30.785
if there was a button on your dashboard

0:00:30.785,0:00:37.128
that you could push, and the car would get you home safely?

0:00:37.128,0:00:39.421
Now, that's been the promise of the self-driving car,

0:00:39.421,0:00:42.048
the autonomous vehicle, and it's been the dream

0:00:42.048,0:00:45.297
since at least 1939, when General Motors showcased

0:00:45.297,0:00:48.599
this idea at their Futurama booth at the World's Fair.

0:00:48.599,0:00:50.542
Now, it's been one of those dreams

0:00:50.542,0:00:54.756
that's always seemed about 20 years in the future.

0:00:54.756,0:00:57.439
Now, two weeks ago, that dream took a step forward,

0:00:57.439,0:01:00.704
when the state of Nevada granted Google's self-driving car

0:01:00.704,0:01:04.304
the very first license for an autonomous vehicle,

0:01:04.304,0:01:06.549
clearly establishing that it's legal for them

0:01:06.549,0:01:08.359
to test it on the roads in Nevada.

0:01:08.359,0:01:12.086
Now, California's considering similar legislation,

0:01:12.086,0:01:14.494
and this would make sure that the autonomous car

0:01:14.494,0:01:17.471
is not one of those things that has to stay in Vegas.

0:01:17.471,0:01:19.567
(Laughter)

0:01:19.567,0:01:23.351
Now, in my lab at Stanford, we've been working on

0:01:23.351,0:01:26.838
autonomous cars too, but with a slightly different spin

0:01:26.838,0:01:31.086
on things. You see, we've been developing robotic race cars,

0:01:31.086,0:01:35.206
cars that can actually push themselves to the very limits

0:01:35.206,0:01:37.446
of physical performance.

0:01:37.446,0:01:40.059
Now, why would we want to do such a thing?

0:01:40.059,0:01:42.159
Well, there's two really good reasons for this.

0:01:42.159,0:01:46.118
First, we believe that before people turn over control

0:01:46.118,0:01:48.952
to an autonomous car, that autonomous car should be

0:01:48.952,0:01:52.206
at least as good as the very best human drivers.

0:01:52.206,0:01:55.511
Now, if you're like me, and the other 70 percent of the population

0:01:55.511,0:01:57.704
who know that we are above-average drivers,

0:01:57.704,0:02:00.879
you understand that's a very high bar.

0:02:00.879,0:02:03.271
There's another reason as well.

0:02:03.271,0:02:06.847
Just like race car drivers can use all of the friction

0:02:06.847,0:02:08.127
between the tire and the road,

0:02:08.127,0:02:11.304
all of the car's capabilities to go as fast as possible,

0:02:11.304,0:02:14.649
we want to use all of those capabilities to avoid

0:02:14.649,0:02:16.237
any accident we can.

0:02:16.237,0:02:18.287
Now, you may push the car to the limits

0:02:18.287,0:02:20.254
not because you're driving too fast,

0:02:20.254,0:02:22.414
but because you've hit an icy patch of road,

0:02:22.414,0:02:24.118
conditions have changed.

0:02:24.118,0:02:26.879
In those situations, we want a car

0:02:26.879,0:02:30.599
that is capable enough to avoid any accident

0:02:30.599,0:02:33.277
that can physically be avoided.

0:02:33.277,0:02:37.544
I must confess, there's kind of a third motivation as well.

0:02:37.544,0:02:39.800
You see, I have a passion for racing.

0:02:39.800,0:02:42.564
In the past, I've been a race car owner,

0:02:42.564,0:02:45.119
a crew chief and a driving coach,

0:02:45.119,0:02:48.974
although maybe not at the level that you're currently expecting.

0:02:48.974,0:02:51.678
One of the things that we've developed in the lab --

0:02:51.678,0:02:53.382
we've developed several vehicles --

0:02:53.382,0:02:55.617
is what we believe is the world's first

0:02:55.617,0:02:57.982
autonomously drifting car.

0:02:57.982,0:03:00.495
It's another one of those categories

0:03:00.495,0:03:02.918
where maybe there's not a lot of competition.

0:03:02.918,0:03:04.326
(Laughter)

0:03:04.326,0:03:08.148
But this is P1. It's an entirely student-built electric vehicle,

0:03:08.148,0:03:10.226
which through using its rear-wheel drive

0:03:10.226,0:03:11.791
and front-wheel steer-by-wire

0:03:11.791,0:03:13.858
can drift around corners.

0:03:13.858,0:03:16.058
It can get sideways like a rally car driver,

0:03:16.058,0:03:17.773
always able to take the tightest curve,

0:03:17.773,0:03:21.077
even on slippery, changing surfaces,

0:03:21.077,0:03:22.693
never spinning out.

0:03:22.693,0:03:25.061
We've also worked with Volkswagen Oracle,

0:03:25.061,0:03:28.485
on Shelley, an autonomous race car that has raced

0:03:28.485,0:03:31.555
at 150 miles an hour through the Bonneville Salt Flats,

0:03:31.555,0:03:36.026
gone around Thunderhill Raceway Park in the sun,

0:03:36.026,0:03:38.665
the wind and the rain,

0:03:38.665,0:03:43.683
and navigated the 153 turns and 12.4 miles

0:03:43.683,0:03:45.245
of the Pikes Peak Hill Climb route

0:03:45.245,0:03:48.718
in Colorado with nobody at the wheel.

0:03:48.718,0:03:50.166
(Laughter)

0:03:50.166,0:03:55.732
(Applause)

0:03:55.732,0:03:59.011
I guess it goes without saying that we've had a lot of fun

0:03:59.011,0:04:00.315
doing this.

0:04:00.315,0:04:03.972
But in fact, there's something else that we've developed

0:04:03.972,0:04:07.027
in the process of developing these autonomous cars.

0:04:07.027,0:04:10.898
We have developed a tremendous appreciation

0:04:10.898,0:04:14.715
for the capabilities of human race car drivers.

0:04:14.715,0:04:19.060
As we've looked at the question of how well do these cars perform,

0:04:19.060,0:04:22.339
we wanted to compare them to our human counterparts.

0:04:22.339,0:04:28.019
And we discovered their human counterparts are amazing.

0:04:28.019,0:04:32.042
Now, we can take a map of a race track,

0:04:32.042,0:04:34.412
we can take a mathematical model of a car,

0:04:34.412,0:04:37.315
and with some iteration, we can actually find

0:04:37.315,0:04:38.940
the fastest way around that track.

0:04:38.940,0:04:41.473
We line that up with data that we record

0:04:41.473,0:04:42.906
from a professional driver,

0:04:42.906,0:04:47.013
and the resemblance is absolutely remarkable.

0:04:47.013,0:04:50.225
Yes, there are subtle differences here,

0:04:50.225,0:04:53.352
but the human race car driver is able to go out

0:04:53.352,0:04:55.687
and drive an amazingly fast line,

0:04:55.687,0:04:58.017
without the benefit of an algorithm that compares

0:04:58.017,0:05:00.625
the trade-off between going as fast as possible

0:05:00.625,0:05:02.662
in this corner, and shaving a little bit of time

0:05:02.662,0:05:04.564
off of the straight over here.

0:05:04.564,0:05:08.021
Not only that, they're able to do it lap

0:05:08.021,0:05:10.396
after lap after lap.

0:05:10.396,0:05:13.308
They're able to go out and consistently do this,

0:05:13.308,0:05:17.436
pushing the car to the limits every single time.

0:05:17.436,0:05:20.605
It's extraordinary to watch.

0:05:20.605,0:05:22.671
You put them in a new car,

0:05:22.671,0:05:26.573
and after a few laps, they've found the fastest line in that car,

0:05:26.573,0:05:30.450
and they're off to the races.

0:05:30.450,0:05:31.596
It really makes you think,

0:05:31.596,0:05:36.467
we'd love to know what's going on inside their brain.

0:05:36.467,0:05:41.008
So as researchers, that's what we decided to find out.

0:05:41.008,0:05:42.820
We decided to instrument not only the car,

0:05:42.820,0:05:45.315
but also the race car driver,

0:05:45.315,0:05:48.084
to try to get a glimpse into what was going on

0:05:48.084,0:05:50.270
in their head as they were doing this.

0:05:50.270,0:05:54.220
Now, this is Dr. Lene Harbott applying electrodes

0:05:54.220,0:05:55.452
to the head of John Morton.

0:05:55.452,0:05:58.441
John Morton is a former Can-Am and IMSA driver,

0:05:58.441,0:06:00.241
who's also a class champion at Le Mans.

0:06:00.241,0:06:03.737
Fantastic driver, and very willing to put up with graduate students

0:06:03.737,0:06:05.592
and this sort of research.

0:06:05.592,0:06:08.264
She's putting electrodes on his head

0:06:08.264,0:06:10.376
so that we can monitor the electrical activity

0:06:10.376,0:06:13.208
in John's brain as he races around the track.

0:06:13.208,0:06:16.403
Now, clearly we're not going to put a couple of electrodes on his head

0:06:16.403,0:06:19.673
and understand exactly what all of his thoughts are on the track.

0:06:19.673,0:06:23.080
However, neuroscientists have identified certain patterns

0:06:23.080,0:06:26.841
that let us tease out some very important aspects of this.

0:06:26.841,0:06:28.688
For instance, the resting brain

0:06:28.688,0:06:30.843
tends to generate a lot of alpha waves.

0:06:30.843,0:06:34.595
In contrast, theta waves are associated with

0:06:34.595,0:06:37.779
a lot of cognitive activity, like visual processing,

0:06:37.779,0:06:40.827
things where the driver is thinking quite a bit.

0:06:40.827,0:06:42.490
Now, we can measure this,

0:06:42.490,0:06:44.475
and we can look at the relative power

0:06:44.475,0:06:46.675
between the theta waves and the alpha waves.

0:06:46.675,0:06:49.117
This gives us a measure of mental workload,

0:06:49.117,0:06:52.684
how much the driver is actually challenged cognitively

0:06:52.684,0:06:54.470
at any point along the track.

0:06:54.470,0:06:57.412
Now, we wanted to see if we could actually record this

0:06:57.412,0:07:00.450
on the track, so we headed down south to Laguna Seca.

0:07:00.450,0:07:02.466
Laguna Seca is a legendary raceway

0:07:02.466,0:07:04.767
about halfway between Salinas and Monterey.

0:07:04.767,0:07:06.854
It has a curve there called the Corkscrew.

0:07:06.854,0:07:09.571
Now, the Corkscrew is a chicane, followed by a quick

0:07:09.571,0:07:12.317
right-handed turn as the road drops three stories.

0:07:12.317,0:07:16.083
Now, the strategy for driving this as explained to me was,

0:07:16.083,0:07:18.105
you aim for the bush in the distance,

0:07:18.105,0:07:21.130
and as the road falls away, you realize it was actually the top of a tree.

0:07:21.130,0:07:24.434
All right, so thanks to the Revs Program at Stanford,

0:07:24.434,0:07:25.907
we were able to take John there

0:07:25.907,0:07:26.871
and put him behind the wheel

0:07:26.871,0:07:29.310
of a 1960 Porsche Abarth Carrera.

0:07:29.310,0:07:33.008
Life is way too short for boring cars.

0:07:33.008,0:07:34.767
So, here you see John on the track,

0:07:34.767,0:07:36.951
he's going up the hill -- Oh! Somebody liked that --

0:07:36.951,0:07:39.416
and you can see, actually, his mental workload

0:07:39.416,0:07:41.569
-- measuring here in the red bar --

0:07:41.569,0:07:43.912
you can see his actions as he approaches.

0:07:43.912,0:07:47.142
Now watch, he has to downshift.

0:07:47.142,0:07:47.903
And then he has to turn left.

0:07:47.903,0:07:51.896
Look for the tree, and down.

0:07:51.896,0:07:54.734
Not surprisingly, you can see this is a pretty challenging task.

0:07:54.734,0:07:57.710
You can see his mental workload spike as he goes through this,

0:07:57.710,0:07:59.774
as you would expect with something that requires

0:07:59.774,0:08:02.583
this level of complexity.

0:08:02.583,0:08:05.999
But what's really interesting is to look at areas of the track

0:08:05.999,0:08:08.845
where his mental workload doesn't increase.

0:08:08.845,0:08:10.340
I'm going to take you around now

0:08:10.340,0:08:11.429
to the other side of the track.

0:08:11.429,0:08:13.765
Turn three. And John's going to go into that corner

0:08:13.765,0:08:16.316
and the rear end of the car is going to begin to slide out.

0:08:16.316,0:08:18.333
He's going to have to correct for that with steering.

0:08:18.333,0:08:20.564
So watch as John does this here.

0:08:20.564,0:08:22.886
Watch the mental workload, and watch the steering.

0:08:22.886,0:08:26.558
The car begins to slide out, dramatic maneuver to correct it,

0:08:26.558,0:08:30.081
and no change whatsoever in the mental workload.

0:08:30.081,0:08:32.913
Not a challenging task.

0:08:32.913,0:08:36.113
In fact, entirely reflexive.

0:08:36.113,0:08:39.756
Now, our data processing on this is still preliminary,

0:08:39.756,0:08:42.428
but it really seems that these phenomenal feats

0:08:42.428,0:08:44.038
that the race car drivers are performing

0:08:44.038,0:08:45.948
are instinctive.

0:08:45.948,0:08:49.338
They are things that they have simply learned to do.

0:08:49.338,0:08:51.620
It requires very little mental workload

0:08:51.620,0:08:54.444
for them to perform these amazing feats.

0:08:54.444,0:08:57.579
And their actions are fantastic.

0:08:57.579,0:09:00.190
This is exactly what you want to do on the steering wheel

0:09:00.190,0:09:03.527
to catch the car in this situation.

0:09:03.527,0:09:06.972
Now, this has given us tremendous insight

0:09:06.972,0:09:10.094
and inspiration for our own autonomous vehicles.

0:09:10.094,0:09:12.022
We've started to ask the question:

0:09:12.022,0:09:14.275
Can we make them a little less algorithmic

0:09:14.275,0:09:16.724
and a little more intuitive?

0:09:16.724,0:09:19.005
Can we take this reflexive action

0:09:19.005,0:09:21.292
that we see from the very best race car drivers,

0:09:21.292,0:09:22.941
introduce it to our cars,

0:09:22.941,0:09:24.925
and maybe even into a system that could

0:09:24.925,0:09:26.893
get onto your car in the future?

0:09:26.893,0:09:28.504
That would take us a long step

0:09:28.504,0:09:31.013
along the road to autonomous vehicles

0:09:31.013,0:09:32.925
that drive as well as the best humans.

0:09:32.925,0:09:36.365
But it's made us think a little bit more deeply as well.

0:09:36.365,0:09:39.333
Do we want something more from our car

0:09:39.333,0:09:41.173
than to simply be a chauffeur?

0:09:41.173,0:09:45.408
Do we want our car to perhaps be a partner, a coach,

0:09:45.408,0:09:48.495
someone that can use their understanding of the situation

0:09:48.495,0:09:52.751
to help us reach our potential?

0:09:52.751,0:09:55.024
Can, in fact, the technology not simply replace humans,

0:09:55.024,0:09:59.599
but allow us to reach the level of reflex and intuition

0:09:59.599,0:10:03.024
that we're all capable of?

0:10:03.024,0:10:04.947
So, as we move forward into this technological future,

0:10:04.947,0:10:07.768
I want you to just pause and think of that for a moment.

0:10:07.768,0:10:11.543
What is the ideal balance of human and machine?

0:10:11.543,0:10:13.252
And as we think about that,

0:10:13.252,0:10:14.983
let's take inspiration

0:10:14.983,0:10:18.312
from the absolutely amazing capabilities

0:10:18.312,0:10:21.128
of the human body and the human mind.

0:10:21.128,0:10:22.864
Thank you.

0:10:22.864,0:10:27.468
(Applause)