Autonomous vehicles will only be able to avoid one third of accidents

Most car accidents are due to human error. Even so, self-driving vehicles would only reduce these accidents by one third. This is the conclusion reached by a study carried out by the Insurance Institute for Highway Safety (IIHS) in the United States.

What seemed like science fiction years ago is closer to becoming reality, and the main car makers, as well as major tech companies, are joining forces to take the lead in the autonomous car race. However, the future might not be as idyllic as it has been described, since self-driving vehicles will only be able to prevent one third of all the accidents that currently happen due to drivers’ errors, according to a report by the Insurance Institute for Highway Safety (IIHS) from the United States.

The IIHS, one of the most respected US institutions regarding transportation safety, questions the optimistic figures offered by the automotive sector, in an effort to promote autonomous driving. According to the IIHS, in order for this technology to prevent most accidents, the programming of the vehicle must focus on safety, instead of speed and comfort.

In order to carry out this study, the IIHS  —an independent entity, not to be confused with NHTSA (National Highway Traffic Safety Administration), the federal agency responsible for highway safety— has analyzed more than 5.000 police reports regarding accidents.

The conclusion is that slightly more than 9 out of 10 accidents that happen in the United States are due to drivers’ errors, which resulted into more than 36.000 deaths last year.

Even though this autonomous technology has been presented in the past few years as a guarantee to prevent most of these accidents, only one third of them would be avoided, because deliberate decisions by drivers, such as speed or illegal maneuvers, cause up to 40% of the crashes, according to the available data.

The main author of this study, Alexandra Mueller, scientific and researcher at the IIHS, states that “making autonomous cars that drive as well as humans do is a big challenge”. Therefore, she emphasized that “they need to be better -these vehicles- in order to make all the promises we have heard become a reality”.

The more than 5.000 analyzed accidents were divided into five main categories. In the first one were included accidents caused by perception and sense errors, such as distractions, poor visibility or the inability to predict danger.

The second one was related to predicting errors, such as poor estimation of the speed of other vehicles. Two other categories were decision and planning errors, such as driving too fast or too slow, and execution and performance, when drivers do not act properly, such as swerving. Finally, the fifth category included alcohol or drug use or falling asleep while driving, or, simply stated, driver incapacity.  Researchers also included preventable accidents, like the one caused by the blow-out of a tire.

With these five categories, researchers considered that autonomous vehicles prevent accidents caused by perception errors or the incapacity of the driver, because cameras and sensors can detect possible dangers better than humans. However, accidents caused by perception errors only account for 24% of the total, and those caused by driver incapacity barely 10%. This is why researchers concluded that autonomous vehicles could only prevent one third of the accidents.

In spite of the conclusions of this report, the European Union predicts that in 2025 this market will generate a profit of 620.000 million euros for the European automobile industry and 180.000 million euros for the electronics sector.