Self driving cars: What is an autonomous vehicle and how do they work?

Self-driving cars have become one of the most hot-button topics in the car industry in the past decade. Everyone from General Motors to Google is involved in trying to remove the human being from the driving process.

Depending on who you believe, self-driving cars are already available, several years away from being fully autonomous and safe and there are some who believe the concept is flawed and will be nearly impossible to achieve.

So what are self-driving cars? When was the first self-driving car invented? And how quickly is the technology changing? We’ll answer these questions and more in this self-driving car explainer.

Read More About Tesla Model S

Autonomous driving tech explained Will fully autonomous cars ever happen? Not anytime soon, according to Volvo boss Self-driving cars will be the end of us | Opinion

What is an autonomous vehicle?

Self-driving cars, also known as autonomous cars or driverless cars, are - as the name implies - vehicles that can be programmed and built to drive themselves without any human assistance or intervention.

Before we go too far, let’s take a look at the official definition of autonomous vehicles as classified by the Society of Automotive Engineers (SAE), these are:

Level 0 - No automation at all but can provide assistance in special cases; e.g. autonomous emergency braking.

Level 1 - Very light computer intervention, such as adaptive cruise control or lane keeping assist.

Level 2 - Mild automation that requires human oversight, e.g. adaptive cruise control and lane keeping assist working at the same time. 

Level 3 -  Can self-drive but requires intervention in emergency conditions. This is where most current autonomous vehicle tech is, such as Audi’s ‘Traffic Jam Assist’, Nissan’s ‘ProPilot’ and Tesla’s ‘Autopilot’.

Level 4 - A highly autonomous vehicle that no longer requires a human to interact with the controls for almost all scenarios. In fact, this is the level where the SAE believes carmakers can get rid of steering wheels and pedals.

Level 5 - A completely autonomous vehicle that can drive everywhere, in all conditions, without any human input at all.

Do they exist already?

Yes and no. There are many cars on the road now with semi-autonomous capability, this can include functions as now-commonplace as adaptive cruise control and lane-keeping assistance.

When these and other systems work in unison a car can begin to drive itself and we’ve seen examples of this with the likes of Tesla’s Autopilot and Audi’s Traffic Jam Assist functionality. 

But there are none currently commercially available that meet the definition of a truly autonomous vehicle, because that is a car that can drive in any conditions, in any scenario, without human intervention in any way, and can even remove the steering wheel or pedals.

For example, a fully-autonomous vehicle in the technical sense, is NASA’s Mars rover, Perseverance. It can traverse the surface of Mars without any assistance from humans on Earth.

Waymo operates a driverless taxi fleet (also known as robo-taxis) in San Francisco and Phoenix. (Image: Waymo)

But, obviously, there’s a big difference between driving along the empty surface of Mars and navigating a busy Australian street with dozens of other cars, pedestrians, cyclists, buses and other obstacles, which is why a true-driverless car is not yet available to buy here (at the time of publication).

The closest is arguably the Waymo One, a fleet of highly modified Jaguar I-Pace SUVs developed by Google’s self-driving spin-off, Waymo, and operating a driverless taxi fleet (also known as robo-taxis) in San Francisco and Phoenix.

The Waymo One has no driver and instead relies on an array of sensors and cameras to handle the driving and navigation, so can be considered at Level 4 autonomy, but is not available for the public to buy yet as it’s still in the testing phase. 

How do self-driving vehicles work?

That’s not a simple question to answer as different companies use different methods.

What most have in common is a variety of technology, with radar, lasers, cameras and sonar all used to detect the surroundings. Each company uses software to process that data and then send signals to the various controls around the car to then drive the vehicle. 

Obviously, ensuring the various sensors pick up all the data and then the software processes it correctly is the major challenge for every company trying to develop truly autonomous cars, as making a mistake can have fatal consequences. 

The Waymo One has no driver and relies on an array of sensors and cameras to handle the driving and navigation. (Image: Waymo)

What is the history of self-driving cars?

Driverless cars date back to the 1920s, when some companies developed radio-controlled vehicles for demonstration purposes.

But the real catalyst for the current wave of autonomous vehicles was a US Defence Advanced Research Projects Agency (DARPA) funded program in the 1980s.

It saw a team from the Carnegie Mellon University use lidar and autonomous robotic control to drive a car without human input. But it was far from the only program of its kind, with similar efforts in Europe. 

Since then the technology accelerated through the 1990s and 2000s, especially with the DARPA Grand Challenge - a US$1 million prize for a team to build an autonomous car that could navigate a 150-mile (240km) course across the Mojave Desert.

In the last 15 years many companies have invested heavily into self-driving car technology including Audi, BMW, Ford, General Motors, Mercedes-Benz, Nissan, Tesla, Toyota and Volvo, as well as Google/Waymo, Uber and even Apple.

Autonomous driving technology has accelerated through the 1990s and 2000s. (Image: Waymo)

What are some current examples of self-driving cars?

As mentioned earlier, no commercially available car can truly claim to be autonomous above Level 3, but most modern cars have some level of autonomous technology, such as emergency braking and lane keeping assist.

The combination of these types of active safety systems that are able to accelerate and decelerate the car, maintain its position in a lane and detect potential collisions are becoming increasingly common, ranging from models like the Tesla Model 3 to the Nissan Qashqai.

What is the future of self-driving cars?

As the technology improves and evolves it will bring down the cost associated with driverless cars and they will likely become more common.

We’ve already seen technology like autonomous emergency braking, adaptive cruise control and lane keeping assist move from being a luxury car item to something mainstream in the past decade.

The obvious next step for the industry is to move to Level 4 and 5 vehicles and have the option to remove the steering wheel and pedals.

Countless concept cars have been launched in recent years either without controls or with wheels and pedals that retract away, but it’s unlikely you’ll find this in a showroom in the near-term.

Tesla boss Elon Musk claimed in 2019 that “within two years” the company would be able to make cars without steering wheels and pedals and launch a fleet of robo-taxis, but has not delivered on either commitment at the time of publication (May 2023).

The other challenge for the industry is legislation, with governments around the world hesitant to allow vehicles on the road without any controls for the driver.

So, even if the car industry is able to develop a safe Level 4 or 5 vehicle, it may not be legal to drive on the road.

Most modern cars have some level of autonomous technology, such as emergency braking and lane keeping assist.

What are the pros and cons of self-driving cars?

While there are those who believe autonomous cars are the answer to a zero road toll, others will claim the technology will never get to such a level and it will be commercially available on a widespread scale.

Technology giant Google is definitely in the ‘pro’ camp, having begun investigating driverless cars back in 2009 and now operating Waymo as a stand-alone business.

A big part of this support seemingly comes from co-founder Sergey Brin, who publicly stated his view that he believes technology will help save lives.

“I expect that self-driving cars are going to be far safer than human-driven cars,” Mr Brin said back in 2012. “'Self-driving cars do not run red lights.”

However, in the real-world, computers are not infallible and in 2022 the US National Highway Traffic Safety Administration released data showing Tesla vehicles have been involved in more than 270 accidents linked to its Autopilot system.

Tesla claims to have one of the most advanced systems on the road with its ‘Autopilot’ (even though it is similar to offerings from other brands) but does warn drivers that the system is an ‘assistant’ after being linked to multiple fatal accidents.

The Waymo One is not available for the public to buy yet as it’s still in the testing phase. (Image: Waymo)

According to the company’s website, the system in the Model 3 and Y “enables your car to steer, accelerate and brake automatically within its lane. Enhanced Autopilot and Full Self-Driving capability introduce additional features and improve existing functionality to make your car more capable over time.”

It’s worth noting that Tesla has been sued multiple times (including a recently-launched class-action by shareholders) over false claims by Musk around the timeframe and capability of full self-driving technology.

Apple co-founder, Steve Wozniak, has been a vocal critic of Tesla and its claims about its Autopilot system.

Wozniak bought a Tesla Model S in 2013 and says the company’s chief, Elon Musk, told him it would be capable of driving across the US autonomously by 2016.

However, in a recent interview with CNN Wozniak made it clear he does not believe that Autopilot is capable of that anytime soon.

“I actually believed those things, and it’s not even close to reality. And boy, if you want a study of AI gone wrong and taking a lot of claims and trying to kill you every chance it can, get a Tesla,” Wozniak said.

Tesla has been sued multiple times over false claims by Musk around the timeframe and capability of full self-driving technology.

It’s not hyperbole from Wozniak about self-driving technology being a matter of life and death, as the tragic case of Elaine Herzberg demonstrates.

Herzberg was crossing a road in March 2018 in Arizona, pushing her bicycle, when she was hit by a Volvo developed by Uber as part of its autonomous project.

The ‘driver’ of the car, Rafaela Vasquez, is due to go on trial in June after being charged with negligent homicide.

Authorities are accusing Vasquez of watching a TV show on her phone while behind the wheel, while her lawyers argue she was using her work phone.

Notably, authorities declined to charge Uber over Herberg’s death despite the car using its self-driving technology, and instead have focused on Vasquez as she was the official operator of the vehicle at the time.

Also worth noting that since the accident and its negative publicity, Uber sold its autonomous vehicle operation to a third-party business, but one headed by Uber’s former CTO.

Tesla claims to have one of the most advanced systems on the road with its ‘Autopilot.’

While less serious, a fleet of 12 Waymo Jaguar’s ended up ‘confused’ and caused a traffic jam in Phoenix in April 2023, which came shortly after five examples got stuck in San Francisco fog earlier in the year, which highlights the fallibility of the concept; at least at this stage.

Which highlights the so-called ‘Trolley Dilemma’, the ethical question of what to do in the event that a trolley is on a collision course and the driver has to choose between killing five people or one person.

This is something carmakers will have to program into their autonomous vehicles, which potentially leaves car companies, and not the driver, responsible for any fatalities.

This becomes particularly important with Level 4 and 5 cars as they could have no controls, therefore removing the driver from responsibility.

In addition to the ethical conundrum, the other major question mark over fully autonomous vehicles may be whether or not it is technically achievable even in the long-term.

John Krafcik, then-head of Waymo, told a CNET interview in 2018 that Level 4 and 5 is a long-term challenge - and one that may not be achievable at all. 

Countless concept cars have been launched in recent years, but it’s unlikely you’ll find this in a showroom in the near-term.

“Autonomy always will have some constraints,” he said. 

“It's really, really hard," Krafcik said. “You don't know what you don't know until you're actually in there and trying to do things.”

Another potential downside to widespread adoption of autonomous cars is mass-unemployment.

While self-driving technology is being sold as a way to make up for a shortfall in the trucking industry and taxis, ultimately it could become another sector of society where automation replaces humans and leads to widespread layoffs.

On the flipside, some car brands have highlighted the positive impact that self-driving cars could have on society.

For example, pro-driverless car members of the industry have suggested it will enable people who cannot drive for a variety of reasons - be it age or disability - to have more freedom with robo-taxis.

One thing that is certain is that carmakers will - for better or worse - continue to develop self-driving cars, but where the technology ends up and how it is received by society - both motorists and legislators - remains to be seen.