Dec. 8, 2022 – What are the cars of the future, and what impact could they have on public health?
Drivers around Silicon Valley have glanced out their car window to see the vehicle cruising alongside them has no one in the driver's seat. Tech companies in California have been testing their new self-driving vehicles, and similar experiments are hitting the road elsewhere as the new technology moves onto streets.
Capable of sensing the environment and moving with little or no human input, these new autonomous ground vehicles are already navigating traffic on public roads. They combine an assortment of cameras and sensors such as radar, sonar, and GPS.
But are a suite of sensors and an algorithm ready to safely take the wheel on the road with us?
Police reported more than 5.2 million motor vehicle crashes in 2020, according to the U.S. Department of Transportation, which resulted in 2 million injuries and more than 25,500 deaths.
Advocates for self-driving vehicles say the biggest cause of most traffic accidents is human error, so taking the driver out of the equation with self-driving vehicles could lead to fewer injuries and deaths. As a future technology, self-driving vehicles are predicted to transform the automobile and insurance industries and change how our cities are planned as traffic patterns evolve.
For a couple of years now, drivers in Bentonville, AR, have been sharing the road with autonomous trucks delivering goods for Walmart. And those trucks now no longer have a driver on hand, ready to take over from the computer if anything goes wrong. The trucks are currently operating independently on the open road.
It's not clear what society's tolerance for crashes or injuries caused by autonomous vehicles will be, even if they happen at a much lower rate than human-caused ones.
In Toronto, similar autonomous trucks are delivering supplies for Canadian grocery chain Loblaws. These deliveries focus on the so-called middle mile and move goods from central depots to front-line stores.
New Driverless Delivery Trucks
The stores launching self-driving vehicles suggest this is the best way to roll them out on public roads because these trips are relatively short and predictable. That means the trucks can run the same route over and over, collecting data on traffic patterns and weather, and that can help refine the algorithm for future trips.
As the technology improves, and if projects like the ones in Bentonville and Toronto are successful, the number of self-driving cars and trucks on the roads could multiply quickly – with profound potential effects on public health.
Supporters of the new technology predict less traffic congestion with autonomous vehicles and therefore less stress and even better air quality, with fewer cars on the road creating emissions.
But Andrew Dannenberg, MD, an epidemiologist and professor of urban design and planning at the University of Washington in Seattle, says the reality of what is planned is more nuanced, and a lot will depend on how self-driving vehicles are actually integrated onto our roads.
Who Is the More Dangerous Driver?
Traffic will only be reduced, he says, if vehicles are shared, which is not particularly popular with most drivers currently travelling alone.
Early experiments where a driver was provided to individuals to be chauffeured to imitate the convenience of a self-driving car found that people drove significantly more when they had the convenience of not driving themselves, says Danneberg.
"If it is too convenient, there will be the same or even more congestion on roads."
And the effect on air pollution is probably relatively neutral, he says. Most of the improvements will come from switching to electric vehicles, which is happening already even before self-driving cars are widely adopted.
Wide access to self-driving vehicles could also lead to people being less active, and less healthy, Danneberg says.
"Physical activity is a big part of transport. But will people walk, bike, or take public transit less if these cars are so convenient? Losing support for transit is not in the public health interest," he says.
But could a combination of good public transit and efficient autonomous vehicles fix our traffic problems? That's possible, if the self-driving vehicles are in the majority, and can communicate with each other to cooperate during rush hour, says Edmond Awad, PhD, who studies the interactions between humans and autonomous vehicles at the University of Exeter in the U.K.
A lot will depend on how the algorithms governing new self-driving vehicles are designed, he explains, and how they alter the level of risk that people have gotten used to on the roads.
Teaching the New Algorithms to Drive
There is a general perception that machines are less biased than humans, says Awad, but since the algorithms are trained on human data, they tend to have the same biases. And they could even make them worse, by putting the same flawed algorithm in hundreds of thousands of vehicles.
"If the algorithm makes cars less cautious around cyclists than the average driver, for example, that will change the distribution of risk," he warns. And since we don't generally know what's going on inside that algorithm, it makes it difficult for people to trust the vehicles and know how to behave around them.
Awad studied how people thought about the algorithms governing autonomous vehicles in a project called The Moral Machine. He gave people hypothetical situations where a self-driving car detects a collision that is about to happen. If the vehicle can't avoid the crash, what could it do next? Collide according to trajectory or swerve to save lives? An algorithm that protects human life would be best, but what happens if veering hits someone else? What if the self-driving car is about to hit a bus, and to avoid all those people on public transit means it drives off the road and kills its own passengers? Should the car be programed to never risk its owners? But what about the crowds of people in public spaces? How will programmers decide whose safety to prioritize?
Awad found in his study exploring hypothetical situations that while most people were in favor of self-driving cars that were programmed in a utilitarian way – that is, the car would sacrifice its own passengers to save greater numbers of other people – people in the study also reported they would prefer not to ride in such cars themselves.
Those designing the vehicles will need to balance the competing priorities of individuals and society at large if the technology is ever going to be accepted in public spaces around other drivers, cyclists, and pedestrians who could be put at risk.
And should self-driving cars have their own closed roads to operate on?
Beyond questions about individual safety, though, there is a whole host of questions about the equity of the transition to autonomous vehicles, says Dannenberg. He is involved in several active outreach projects to determine what communities want from autonomous vehicles. He has found that many lower-income or otherwise marginalized communities have little input or interest in the issue.
They are generally not in contact with the politicians and technology leaders who are deciding where, when, and how the vehicles will be rolled out, and so are not high on the priority list, says Dannenberg.
"If the only people who care about this are well-off, that's a big equity issue."
Marginalized communities face larger barriers to adopting the technology, either through price, access to other enabling technologies like smartphones, or simply which neighborhoods are served by autonomous vehicle projects.
Many jobs in transportation will also be replaced by the vehicles, a burden that will fall unequally on those with lower incomes or less education.
Those with disabilities may also face greater barriers, if there is no human on hand to help wheelchair users get in and out of the vehicles they want to travel in.
"It's not automatically great for disabilities," says Dannenberg. That's an issue that can be solved with clever design, but it requires careful thought and comes with a cost.
Dannenberg says the equity issues are getting less attention than they deserve as society prepares to adopt more self-driving vehicles, so more people from different sectors and communities need to get involved in driving this change in a direction we want to go.
One way or another, people are in the driver's seat of this technological advance.
"And we will need ongoing monitoring and evaluation," Dannenberg says.