Probably the biggest of our generation.

- [Narrator] This is a Tesla.

It comes with a feature known as autopilot.

And Elon Musk says in the future,

it'll have a feature called.

- Full self-driving.

- [Narrator] Several other cars being sold today

also come with assisted driving technology

which still requires someone behind the wheel.

But new vehicles that won't have anyone in the driver's seat

are getting closer to production.

Startups Waymo, Cruise, TuSimple, and Aurora,

are already testing driverless tech

on some public roads across America.

- It's gonna move quickly.

We'll start to see this move from prototypes

to actually scaling products (indistinct).

- We see us becoming fully driverless in 2024.

- [Narrator] As they invest billions in R&D

and sign multi-billion dollar deals,

their valuations have soared and TuSimple has gone public.

They're pitching a future where people won't even own cars.

- Younger generations won't even think about owning a car.

- [Narrator] And some jobs will no longer exist.

- We fundamentally believe that every trucker will be able

to retire as a trucker.

- [Narrator] But first of all,

they need to convince governments

and the public that their technology is safe.

- You're limiting your potential

if consumers are not comfortable getting into your vehicle

and taking a ride.

- [Narrator] So could these startups make driving

a thing of the past?

Tesla has been working

on making its vehicles autonomous since around 2016

and said it planned to have self-driving cars

on the road by 2020.

- And sometimes I'm not on time.

But I get it done.

- [Narrator] For now, experts say these cars

have what they call level two autonomy.

- It is a level two system where they're saying

that the driver has to be there, they have to pay attention.

They're counting on them.

- [Narrator] What does that mean?

Well, it's a scale used by auto engineers

and it starts at zero.

- Which means, you know, kind of a 1950s car

with no automation at all.

- [Narrator] And goes up to level five.

- [Tekedra] Basically when the car can do everything,

anywhere.

- You can drop it off in Afghanistan today

and it will be able to operate without maps by itself.

- [Narrator] Instead of releasing vehicles

and working towards autonomy,

these top startups want to roll them out directly

at level four autonomy.

Which is where the car doesn't need a driver

as long as it's preloaded with information

about its surroundings like maps and directions.

- And we just simply think level four is the goal

for any domain.

- [Narrator] Some companies in the US and Asia

like Baidu AutoX and Didi,

say they've reached level four in some vehicles

though none have rolled them out at commercial scale

but some are getting close.

Waymo and Cruise have piloted a fleet of robotaxis

in some ring-fenced areas around the US.

Think ride-hailing without the driver.

Like this, the Waymo One.

- We're the only company

that has a fully autonomous ride-hailing service

available to the public today in the Phoenix metro area.

- [Narrator] In late 2020,

Waymo began removing support drivers

from its vehicles after its cars racked up

some 20 million miles on public roads,

gathering troves of data for its algorithms.

- The car shows up completely empty.

They take a ride from point A to point B.

That's one of the ways you make this real,

is you launch a service in the city

and you see if customers will actually adopt the service

and use it.

- [Narrator] Waymo started as a Google project in 2009,

and now has a $30 billion valuation and Alphabet,

Google's parent, as a majority shareholder.

Cruise has roughly the same valuation as Waymo

and is owned by GM.

With whom it has built an autonomous vehicle from scratch.

It has accrued fewer miles than Waymo, some two million,

in it's tests in San Francisco.

- Miles driven absolutely matters

because ultimately the goal is to put this technology

on public roads,

interacting with other human-driven vehicles.

So you do need that experience at some point to say,

yes, this is not only safe, but it's actually viable.

- [Narrator] Cruise recently got permission

to remove the support driver

but it's yet to carry paying passengers.

It plans to begin production of this vehicle,

the Origin, in 2023.

We spoke with company representatives over the phone,

who said GM's backing is a strategic advantage

but it's not just cars, startups are also eyeing trucks.

- The trucking industry is a massive industry.

It's $4 trillion globally,

800 billion dollars in the US alone.

- [Narrator] That's the market Aurora

and TuSimple are betting on.

Aurora which has received investment from Uber

says it has a handful of test vehicles out

on public roads in the US.

TuSimple has around 60 autonomous trucks circulating

on highways in the Southwest with support drivers.

It was the first driverless company in the US

to go public, getting a roughly $8.5 billion valuation.

- In the US alone, 60,000 driver shortages this year

that will increase to about 160,000 by 2028

based on latest numbers.

So it's a very real problem.

- We don't intend to ever build a car or build a truck.

We work with great companies like Paccar and Volvo

and Toyota, and integrate our technology to their vehicles.

- I think consumer goods can provide, you know,

an easier path.

You're already kind of simplifying the situations

that you would have to deal with.

If you're driving the same pathway often,

you're not gonna have pedestrians and cyclists, you know,

interacting with traffic.

- [Narrator] But their tech needs

to overcome some big challenges

before it can be widely adopted.

- And the more we discover about the challenges,

the more we learn about how long this is gonna take.

And so I think you saw all of industry

start to reset expectations.

- [Narrator] Waymo and others had previously said 2020

was the year that robotaxis would become commonplace

across the US but the coronavirus pandemic

and technical hurdles

have led to forecasts being pushed back.

One issue, these vehicles still don't work well at night

or in bad weather.

- It's not a coincidence that you see companies testing

in fair weather places first.

- When you think about the barriers,

it's really engineering barriers.

It's time, it takes capital, but we'll get there.

- [Narrator] But those are not the only barriers.

- A Honda Civic right there colliding with the Waymo van.

We're being told it was not at fault in this accident.

- [Narrator] The industry has to address safety concerns.

As autonomous cars have been involved

in a number of accidents in recent years.

After a fatal crash in 2018,

where a car's built in automatic emergency braking system

was disabled, Uber had to stop testing autonomous vehicles

in Arizona.

It continued tests in other cities

before selling its unprofitable self-driving car unit

to Aurora at the end of 2020.

All players in this industry know

that they have to deal with a lot of skepticism.

- There's this suggestion that the technology

is presenting new risk or first risk.

The reality is though roads have risk.

- So it's not possible to make anything 100% safe,

whether it's a toothbrush or an airliner.

And so we try to make these systems

really safe.

- [Narrator] A recent study found

that even if autonomous vehicles cause accidents

at half the rate of human drivers,

only 37% of Americans would opt in.

- We've got more forgiveness, we've got more understanding

if that happens or is caused by another human driver.

I don't think we have that same compassion for a robot

that might produce the same result.

- [Narrator] To show that self-driving vehicles

because very few accidents,

Waymo started disclosing all of its crashes

and near misses in Arizona,

a total of 47 episodes over 2019 and 2020.

- Because we thought it was really important

to start to demonstrate

that you insert a fully autonomous vehicle into an ecosystem

and there is still human error around the vehicle.

- Sure, we've got disengagements, we've got miles driven,

we've got the number of crashes

and we can relate that to human drivers,

but that doesn't necessarily make it much easier

to kind of say, this is safe, safer, or safe enough.

So just being able to quantify these things

I think it's still a challenge for the industry.

(gentle music)