It has become day-to-day talk that driverless vehicles can cause accidents that result in fewer deaths and injuries than human-driven vehicles


It has become day-to-day talk that driverless vehicles can cause accidents that result in fewer deaths and injuries than human-driven vehicles. But so far, most comparisons between drivers and automated vehicles have been unbalanced at best and unfair at worst.

The statistics that measure how many accidents occur are hard to argue with: more than 90 percent of cars in the U.S. are thought to be driverless. Eliminating mortal accidents is said to save as many people in two years as the country lost in the entire Vietnam War.

A total of 1 million 182,491 traffic accidents occurred in Turkey in 2016. 185 thousand 128 of these rates are fatal injury traffic accident (TurkStat 2017 report, Highway Traffic Accident Statistics No. 24606)

For me, as a human researcher, there is not enough information to accurately assess whether automation can be lower than human accident rates. The collision rates of driverless vehicles can only be determined by knowing how many collisions did not occur.

For human drivers, is the probability of a collision one per billion or one per trillion?

Assessing the rate at which things are happening is extremely difficult. For example, guessing that you didn't hit someone in the hall today is about how many people were in the hallway and how far you walked there. To determine whether automated vehicles are safer than humans, the researchers need to establish a collision avoidance rate for both humans and newly released driverless vehicles.

Comparing appropriate statistics

Crash statistics for manned cars are compiled from all types of driving situations and on all types of roads. This includes driving through dirt roads and climbing steep slopes in the snow when it rains. However, much of the data on the safety of its driverless cars, often in good weather, comes from Western states of the United States. The multifaceted data was recorded on one-way, multi-lane highways, where the most important task is to keep the car in its lane and not get too close to the vehicle ahead.

When we look at Turkey, we have clearer data: 43.5% of the people who died in traffic accidents in 2016 were drivers, 33.1% were passengers and 23.4% were pedestrians. In 2016, a total of 213,149 defects caused by fatal injuries were caused by the driver, 89.6% by the pedestrian, 8.7% by the pedestrian, 0.8% by the road, 0.5% by the vehicle and 0.4% by the passenger.

Contrary to common belief, at least a maximum in July of accidents happened in the month of January in Turkey

As states allow wider use of driverless vehicles, data on fully automated systems will naturally be more on the agenda . But it's a fact that it will take some time for driverless cars to be able to cover as many areas in a year and as many as human drivers do now…

It's true that driverless cars aren't tired, angry, frustrated or drunk. But they cannot react to uncertain situations with the same skill or expectation as an attentive human driver. In addition, fully automated vehicles have the foresight to avoid potential danger, but rather than literally thinking about possible events down the road, driverless vehicles need to assess the moment at a large rate.

On the automatic vision system, a bus stop full of people can look quite similar to a deserted stop . Indeed, it is difficult for people to decide what action should be taken in an emergency, but drivers can sacrifice themselves for the good of others.

The fact that an automated system has a limited understanding of the world means it will never assess a situation the way a human does. Machines cannot be pre-programmed to handle every conceivable set of events.

New technologies bring new concerns

Some people may simply argue that the promise to reduce the number of injuries and deaths is enough to justify popularising the use of driverless vehicles. I agree that if there is news of a new day tomorrow where a completely indefinitely road doesn't kill or injure anyone, it would be a great thing.

Weather experience, when introduced to new automated systems, is generally assumed to be an increase in the rate of adverse events. However, most accidents in Turkey occur in July, when the weather conditions are very good.

As a result, comparisons between humans and automated vehicles should be made with caution. Since these human-controlled vehicles are likely to remain on the roads for many years and even decades, the likelihood that driverless vehicles could remain on the road should be well calculated. How do people and driverless cars fit together? Who will be the culprit in any collision between them?

To assess driverless cars fairly about how well they meet the promise of improved safety, it is important to make a real comparison of the data being presented. Choosing to replace people with automation has more impact than just swapping. We have time to think a bit more before driverless vehicles cover the roads…