Transcript of YouTube Video: Why You Should Want Driverless Cars On Roads Now

Transcript of YouTube Video: Why You Should Want Driverless Cars On Roads Now

The following is a summary and article by AI based on a transcript of the video "Why You Should Want Driverless Cars On Roads Now". Due to the limitations of AI, please be careful to distinguish the correctness of the content.

Article By AIVideo Transcript
00:00

- All right, I'm about to go for my first ever ride

00:02

in a fully autonomous vehicle.

00:09

Whoa, no driver.

00:14

All right.

00:15

- [Electronic Voice] Good morning, Derek.

00:16

This car is all yours with no one up front.

00:19

- I really like the idea of fully autonomous vehicles,

00:22

but it's weird getting into a car with no driver

00:26

and just trusting the car.

00:28

I'm gonna report back how this ride goes

00:32

and how I feel about it.

00:34

Oh, but full disclosure, this video is sponsored by Waymo.

00:41

(car door clicking)

00:43

(seatbelt clicking)

00:44

Start ride. (car navigation chiming)

00:46

Ha ha!

00:47

- [Electronic Voice] Make sure your seatbelt is fastened.

00:49

(Derek vocalizes) For any questions,

00:50

press the Help button to speak with a rider support agent.

00:53

- Okay. Now let's see where we go.

00:56

It's looking at this car coming here. What's it gonna do?

01:00

Oh, and it pulls in very smoothly behind it.

01:04

No problems. Did not turn into traffic.

01:06

Waited until the cars went, and then it turned.

01:09

I like that.

01:11

I polled YouTube viewers about autonomous vehicles

01:14

and half of you are excited

01:16

and ready for them to be on the roads,

01:18

but over 40% said you thought the technology

01:21

was still over 10 years away.

01:24

And for those people, I have news,

01:26

which is that, well, there is no driver in this car.

01:30

I'm currently inside a fully autonomous vehicle driving

01:34

around a suburb of Phoenix, Arizona.

01:37

Now I get that in some parts of the world,

01:39

like, the roads aren't well enough maintained

01:41

and people don't stay in their lane necessarily,

01:44

and so it'd be very hard for a computer to drive there,

01:46

but at least under good conditions,

01:49

the technology is currently functional.

01:51

(downtempo music)

01:57

(bright comedic music)

01:58

- Now Waymo started out

02:00

as the Google self-driving car project

02:02

with what is possibly one of the cutest cars ever made.

02:08

I am inside the world's first fully autonomous vehicle.

02:12

Back in October 2015, this car went on a public road,

02:17

ridden by Steve Mann,

02:19

who has a disability, he is legally blind,

02:22

but he could get around in this thing,

02:25

which is affectionately known as the Firefly.

02:28

This is such a simple vehicle.

02:30

There's basically nothing in here.

02:32

There's no steering wheel, no dashboard.

02:34

This car is super basic.

02:36

There is no AC, but there is an emergency stop button.

02:40

(button cover clicks)

02:41

That's my favorite button in the car.

02:43

It reminds me of elevators.

02:45

One of the important measures that they had to put

02:47

in automatic elevators was a big red stop button.

02:51

Did you know that before the 1940s,

02:53

almost all elevators had drivers in them?

02:56

And when people started putting in driverless elevators,

03:00

well, the public was very concerned,

03:02

and they didn't wanna ride in those elevators.

03:03

There was one guy who was like,

03:04

"I don't care if I have to walk up 12 flights of stairs

03:07

for the rest of my life.

03:08

I'm not taking that elevator."

03:10

And adoption was slow.

03:11

I mean, they tried to advertise

03:12

to help people understand that it was in fact safe,

03:15

but, ultimately, there was an elevator drivers strike

03:18

in New York City, and that really annoyed people,

03:21

and it helped the adoption of automated elevators.

03:25

If you found a driver in an elevator today,

03:27

you would wonder, "Why are they there?"

03:30

Now you might think an elevator is just so simple,

03:32

I mean, it is effectively one dimensional motion,

03:35

but airplanes are also flown extensively by computers.

03:40

(tense percussive music)

03:42

I saw this particular landing

03:44

where a plane is coming in into Vienna,

03:47

and it's just so foggy

03:48

that the pilots can see almost nothing.

03:50

I mean, this is the view from the cockpit,

03:53

and yet, they make a perfect,

03:55

a textbook landing right on target.

03:58

So how do they do it?

04:00

The answer is the pilots didn't do it.

04:03

It was a CAT II autoland procedure.

04:05

The plane just came in and landed itself essentially.

04:09

Now, of course, the pilots are important

04:11

and they're monitoring all of the instruments and controls,

04:15

but it's actually the plane

04:17

and its computer getting the plane to land appropriately.

04:20

I was surprised to learn that humans are much more likely

04:24

to take manual control and land on sunny days,

04:28

like July 6th, 2013, when Asiana Airlines flight 214

04:32

was on final approach to San Francisco.

04:35

Attempting to manually land the plane,

04:37

the pilot accidentally left the throttle at zero,

04:40

and by the time they realized

04:42

and tried to abort the landing, it was too late.

04:46

The plane crashed into the runway seawall and split in two.

04:51

Three people died in the aftermath of this accident.

04:55

I think the counterintuitive thing

04:57

is that we expect the humans to be better,

05:00

particularly in tough situations,

05:02

but when it comes to airplanes, if it's bad weather,

05:05

you actually want the plane flying itself.

05:09

So the obvious next question is,

05:11

would you want the same thing for cars?

05:13

(oscillating music)

05:15

There are all these different levels of autonomy,

05:17

and everything up to four requires a human driver

05:20

to be responsible and have the wheel at all times.

05:24

In the early days of the Google self-driving car project,

05:27

they had a vehicle that was not yet level four,

05:29

so it still required a human driver.

05:31

They let Google employees borrow the cars,

05:34

but they still had to be in control of the wheel.

05:36

And the volunteers were informed

05:38

that they were responsible for the car at all times

05:41

and that they would be constantly recorded,

05:42

like video recorded, while they were in the car.

05:45

But still, within a short period of time,

05:47

the engineers observed drivers rummaging around

05:50

in their bags or checking phones, putting on makeup,

05:53

or even sleeping in the driver's seat.

05:57

All these drivers were trusting the technology too much,

06:00

which makes almost fully autonomous vehicles

06:03

potentially more dangerous than regular cars,

06:06

I mean, if the driver is distracted

06:07

or not prepared to take over.

06:10

So this is why Waymo decided

06:12

that the only safe way to proceed is with a car

06:14

that has at least level four autonomy.

06:21

This is the depot where the cars go

06:23

when they're not on the road.

06:24

And it's also where people monitor

06:26

all the rides in progress.

06:28

- Yes, so that's where my team sits.

06:30

You see three teams basically here.

06:32

One of the teams is my team of fleet dispatchers,

06:35

so basically making sure

06:36

that all the mission are assigned every day

06:39

and they are completed successfully on the road.

06:41

And then you have the Rider Support team

06:42

that takes the calls.

06:45

(electronic tone chiming)

06:46

- [Beulah] Thank you for calling the Waymo Rider Support.

06:48

This is Beulah, how can I help today?

06:50

- I just completed my ride,

06:51

but I don't wanna get out of the car.

06:53

I just wanna keep driving.

06:54

Is there a way that I can do that?

06:55

- [Beulah] Right now, I don't see a trip started.

06:58

Give me one moment here while I partner with my team, okay?

07:00

- Okay.

07:06

(chuckling) Well, I just left all my stuff in the car.

07:11

I hope it comes back.

07:17

(car humming)

07:18

- I think there's a lot of, still, resistance

07:21

in terms of trusting the vehicle.

07:24

And they ask you like,

07:26

"How does it feel to be in a car without the driver?"

07:29

I was the first person to do public roads,

07:33

fully driverless ride at night.

07:35

I always share the experience with them.

07:36

It takes about two minutes for you to completely forget

07:40

that you are in a driverless vehicle.

07:42

If the system really provides that feeling that you're safe

07:47

and you see a couple of maneuvers,

07:48

in less than two minutes,

07:49

you're talking to whoever is next to you

07:51

and not paying attention to what's happening anymore.

07:56

- Whoa, all right, it doesn't make the indicator sound,

08:00

so I just don't know when it's going to turn,

08:04

but if I was watching the map, I would know.

08:08

I think we have this bias to believe

08:11

that we're better at certain tasks than we actually are,

08:14

like thinking that people are good at driving.

08:17

Surveys show 74% of people believe

08:20

they are above average drivers.

08:23

Think about that.

08:25

In the 20th century,

08:26

60 million people were killed on the road.

08:28

That's basically an extra world war's worth of deaths.

08:32

And we really have no one to blame but ourselves.

08:34

The National Transportation and Safety Board

08:36

has identified human error as the cause of 94% of accidents.

08:41

Most of these errors are impossible for a machine to make.

08:46

Every year when people are backing out

08:48

of driveways or parking spaces,

08:51

in the U.S, up to 200 people are killed,

08:54

and it's frequently older people or children,

08:57

the children of the drivers.

08:59

It's awful.

09:00

And it comes down to the fact

09:01

that we don't have eyes in the back of our head,

09:03

and even the backup cameras still have blind spots.

09:07

But if you have a vehicle that has LiDAR

09:10

and radar and 29 cameras,

09:13

you're just not going to hit them.

09:16

Up here in the very prominent top, there is a 360 LiDAR.

09:21

So it can see all around the car.

09:23

It can see up to 300 meters away with a LiDAR.

09:27

The way the LiDAR works

09:29

is it shoots out invisible laser beams,

09:32

scanning around millions of times a second,

09:35

and then it detects the reflection,

09:37

and how long it takes to come back allows you

09:39

to determine how far it is to that object.

09:42

So what it's doing is like painting a 3D picture

09:45

of the world.

09:47

There are 29 cameras around this vehicle,

09:50

which gives you full 360 vision.

09:53

It gives you close range vision,

09:55

what is right next to the car,

09:56

and also long range vision, going out 500 meters.

10:01

This car could detect a stop sign

10:02

or a pedestrian 500 meters away.

10:05

How many of us have eyesight that is that good?

10:08

There is also a microphone up on top

10:12

to listen to what's happening in the environment,

10:14

and if there are sirens,

10:16

then the car will pull over to the side of the road.

10:19

It's gotta be able to respond to emergency vehicles.

10:22

(light percussive music)

10:31

What I wanna see here is, how does it handle a parking lot

10:36

where there's people driving in unusual ways

10:39

and possibly pedestrians walking around?

10:42

(car humming)

10:46

(tires squeak) Whoa.

10:48

That was a sudden stop.

10:52

The car made a pretty hard stop there.

10:53

I think it saw that guy with a cart coming up

10:56

on a pedestrian crosswalk.

10:57

And one of the interesting things

10:59

that the vehicle is always doing

11:01

is not only seeing where things are

11:03

and where they're going, but also making predictions

11:06

about where they're likely to go.

11:08

So this car doesn't just have one potential future.

11:12

It's constantly imagining,

11:13

"Well, he might cross at the crosswalk,"

11:15

or, "He might keep going," or, "He might turn left,"

11:18

and so it has to be prepared

11:20

for all of those different options.

11:22

And it even weights the options of like how likely it thinks

11:25

that he's gonna go on the crosswalk

11:27

versus go straight versus turn.

11:29

And you can see that with the thickness of the line

11:32

in the little simulated graphic that they have.

11:37

(seatbelt clicking) Phew.

11:38

A few years back, I think a lot of people were talking

11:40

about how autonomous vehicles have to figure out

11:44

who to hit in case of an accident,

11:46

like do the pick the orphan or the nun?

11:49

Should the car hit the motorcyclist with the helmet on

11:51

because his injuries might be less severe

11:53

or should the car hit the motorcyclist

11:55

who does not have a helmet on

11:56

because he did not properly protect himself?

11:59

If cars were programmed

12:00

to hit the motorcyclist with the helmet,

12:02

that would mean that, in a way,

12:03

it would become safer to ride without a helmet.

12:05

(whimsical music)

12:06

But the reality is that 99% of accidents aren't like that.

12:11

Every year, around 1.3 million people are killed

12:15

on the roads, almost all of them due to human error.

12:19

If autonomous cars can reduce these fatalities,

12:22

then the real moral dilemma

12:23

is not getting them on the road sooner

12:25

for fear we haven't worked out exactly how they'll react

12:28

to extremely unlikely hypothetical scenarios.

12:33

I think humans are becoming worse drivers

12:36

because we're just so prone to distraction.

12:38

(light contemplative music)

12:39

Think about the main reasons why cars crash:

12:44

because people are speeding,

12:46

they're under the influence, they're distracted.

12:50

I mean, these sorts of problems,

12:53

an autonomous vehicle would not have.

12:55

You don't get a distracted driver.

12:57

The ultimate question, right, that everyone wants to know,

12:59

that I want to know the answer to is,

13:02

as these vehicles stand, (car electronic beeping)

13:03

are they better than the average human?

13:05

Not than the best human, but just like an average human.

13:08

Like replacing some random car on the street

13:11

with one of these vehicles, does that make the road safer?

13:13

- Yes, it does.

13:14

I think we would never launch a rider-only service

13:19

if we did not meet that base safety framework.

13:23

- If that's true, it means like every vehicle

13:27

that's not on the road is kind of a worse situation,

13:31

do you know what I mean? - We are really working

13:33

really hard to launch this in larger areas

13:36

and new areas too, but we need to have the experience

13:39

to show the regulators why we believe it's safer,

13:42

and, for that, you need to be driving miles,

13:45

a number of miles that you feel comfortable with,

13:48

statistically speaking.

13:50

- These vehicles have way more experience

13:52

than any human driver because they've now accumulated data

13:56

over 20 million miles of driving on public roads.

14:00

If you were an average driver,

14:02

you'd have to drive for a thousand years

14:05

to accumulate that sort of experience.

14:07

And all of that experience can be used

14:09

to train the systems, to fix the software,

14:11

and used across all the vehicles in the fleet.

14:15

In 2019, Waymo released a study of its data,

14:18

over 6.1 million miles of automated driving

14:21

in the Phoenix, Arizona metropolitan area.

14:24

Of the 18 total accidents that occurred during the study,

14:27

none were serious enough

14:29

to expect significant injury or death.

14:31

In Waymo's Safety Report,

14:32

they found some types of accidents

14:34

have been completely eliminated

14:37

by this autonomous driving system,

14:39

like the car doesn't go off the road

14:42

and it doesn't hit stationary objects.

14:44

Humans, humans do those things.

14:47

If you look at the eight significant accidents

14:51

that happened with Waymo vehicles

14:52

over the six million miles of driving,

14:55

all eight of them involve a human driver

14:58

of another vehicle doing something stupid,

15:01

like driving on the wrong side of the road

15:03

or running a red light

15:04

or going through a stop sign or failing to yield

15:07

or going 20 miles per hour over the speed limit.

15:10

There were three incidents involving Waymo vehicles

15:13

and pedestrians, but in all three,

15:17

the Waymo vehicle was stationary and the pedestrian

15:21

or cyclist/skateboarder ran into the vehicle.

15:26

Waymo also takes some of that real-world data

15:29

and they put it into simulations and they tweak it a bit.

15:33

So they try adding like a bicyclist going fast

15:36

or going slow, or they make the car turn faster or slower.

15:39

So they change all these parameters and variables,

15:42

and they see what the software will do.

15:44

And they've trained the software

15:46

on an additional 20 billion miles of driving,

15:49

not on the road but in the simulation,

15:51

so that's a thousand times more experience again.

15:56

The question for me is,

15:58

when is stepping inside an autonomous car

16:01

gonna feel the same as stepping inside an elevator?

16:05

Because I think that time may be coming sooner

16:07

than you think.

16:09

I like the idea of this technology,

16:10

but, honestly, getting in the car,

16:11

I wasn't quite sure how I would feel.

16:14

I was a little bit uncertain,

16:16

but once I saw it just handles so confidently...

16:19

Like driving is one of those things

16:20

that I feel like you can't hide

16:21

whether you're a good or bad driver.

16:22

It's just like, "Oh, what's gonna happen

16:24

when there's a parked car or a cyclist or a pedestrian,"

16:27

and it just sort of handles all those situations

16:30

with such confidence and ease

16:32

that I think I stopped thinking about it.

16:34

- After they pass a couple of your mental tests,

16:36

you're like, "I'm good."

16:38

I think, again, "I'll be okay." (chuckling)

16:40

- Yeah, yeah, I felt the same way.

16:41

I think a lot of people miss the bigger implications

16:44

of what is achievable

16:45

once fully autonomous driving is commonplace.

16:48

Riders with disabilities, seniors,

16:50

and the blind can get around more easily.

16:52

Transportation will get cheaper.

16:54

Think of all the wasted value in the cars

16:56

that spend over 95% of their time parked.

17:00

We can regain a bunch of time and feel happier

17:03

because commuting and being stuck in traffic sucks.

17:07

We can reduce traffic

17:09

because vehicles will have better awareness of each other.

17:11

You can imagine one day

17:13

when all the cars are fully autonomous,

17:14

they can execute a beautiful ballet driving together,

17:18

and when that time comes, we can eliminate parking lots

17:21

and add green spaces to our cities.

17:23

And, most importantly,

17:24

widespread adoption of autonomous cars

17:26

could prevent tens of thousands of fatalities

17:29

in the U.S. alone.

17:30

When do you think that this will be a reality,

17:33

but clearly it's coming,

17:34

but when? - Yeah.

17:35

If you're talking about big cities,

17:37

I'm hoping the next five years will be really game changing.

17:40

- [Derek] I'm excited to see it.

17:41

- Me too, believe me. Believe me.

17:43

I just don't want to commute to work anymore.

17:45

I would love to sit in the backseat,

17:46

do my work on the way there and on the way back. (chuckling)

17:49

- [Derek] Or do something fun, watch a movie.

17:51

- Or sleep, yeah. (chuckling)

17:52

(futuristic tones beeping)

17:53

(frequency tone whistling)