It goes without saying that self-driving cars are held to incredibly high safety standards, but what we often forget is that the real standard is often the dangerous driving of humans. Now, a new study shows that in most scenarios, self-driving cars cause fewer accidents than humans.
One of the main arguments for switching to self-driving cars is the prospect of eliminating human error from driving. Given that more In the United States, 40,000 people die in traffic accidents every year.Even small improvements in safety can make a big difference.
However, self-driving cars have been involved in numerous accidents in recent years, raising questions about their safety and causing larger-scale accidents. companies like cruises To scale back their ambitions.
But now researchers have analyzed thousands of accident reports from incidents involving both self-driving cars and human drivers. Their results were published as follows: Nature Communications, This suggests that self-driving cars are actually safer than humans in most situations.
The University of Central Florida team focused their research on California, where most autonomous vehicle testing is taking place. They collected 2,100 accident reports involving self-driving cars from databases maintained by the National Highway Traffic Safety Administration, the California Department of Motor Vehicles, and news reports.
They then compared this to 35,000 accident reports involving human drivers collected by the California Highway Patrol. The team used an approach called “matched case control analysis” to try to find pairs of crashes involving humans and self-driving cars with very similar characteristics.
This allows us to control for all other variables that may contribute to a crash and examine the impact that “the driver” has on the likelihood of a crash occurring. The team found 548 such matches, compared the two groups, and found that self-driving cars were safer than human drivers in most of the accident scenarios they looked at.
However, there are some important caveats. Researchers also found that self-driving cars were more than five times more likely to be involved in an accident at dawn or dusk and almost twice as likely when turning.
The former is likely due to limitations of the imaging sensor, while J. Christian Gerdes of Stanford University said IEEE spectrum The turning problem is probably due to our limited ability to predict the actions of other drivers.
But self-driving cars also had some positive aspects. They were roughly half as likely to be involved in a rear-end accident and only one in five as likely to be involved in a side collision.
The researchers also found that self-driving cars are only about one-third as likely to crash in rain or fog as they are for human drivers, which they believe is because the vehicles rely on radar sensors that are not significantly affected by bad weather. I did.
How readable these results are is debatable. The authors acknowledge that limited data on autonomous vehicle crashes limits the scope of their findings. Missy Cummings of George Mason University said new scientist Accident reports from self-driving companies are often biased, trying to shift blame to human drivers even when the facts do not support it.
Nonetheless, the study is an important first step in quantifying the potential safety benefits of autonomous vehicle technology and highlights several important areas where progress is still needed. Policymakers need a clear look at the numbers to make informed decisions about when and where to deploy this technology.
Image source: gibblesmash asdf / Unsplash
