What Do Self-Driving Cars Mean for Road Safety?
January 23, 2019

One of the most important safety features of any vehicle is the driver; experience level, judgment, and avoiding driving under the influence all help avoid accidents. The proliferation of driverless vehicles and autonomous vehicles has caused many Americans to wonder what this means for the safety of U.S. roads, especially after an autonomous Uber vehicle struck and killed a 49-year-old woman in Tempe, Arizona in March of 2018.

While a driver can be an important safety feature, the driver is also often the cause of traffic accidents. Driver error is one of the most commonly cited reasons for traffic accidents all over the country, and self-driving cars may eventually take human error out of the road safety equation entirely.

Constant, Real-Time Accident Prevention Updates

Drivers learn from experience, and while other drivers may teach them the fundamentals of driving, every driver ultimately builds muscle memory and driving habits with experience. It is impossible for one driver to absorb the experiences of another, but an entire fleet of self-driving cars can update all their individual operations seamlessly thanks to the data collected by a single unit.

Self-driving vehicles feature machine learning programs that can analyze human driver actions as well as accident data from many different types of vehicles.

For example, a self-driving vehicle could assign different profiles to different types of vehicles and assign different levels of risk to each type. Some self-driving programs could eventually learn to detect the presence of a nearby intoxicated or dangerous driver and help avoid a serious accident.

The safety and reliability of a self-driving vehicle ultimately comes down to the software running the driving processes. As a vehicle gathers data from driving experiences – including accidents – the vehicle’s manufacturer can apply the gathered data to every other model that uses the same software, essentially allowing all vehicles that share the same software to learn from one another in what is almost real-time.

Improving Road Safety With Data Collection

When self-driving vehicles absorb enough collective experiences to completely avoid accidents in most possible road situations, the data gathered could then apply toward making the roads safer in other ways. For example, data collection could allow road maintenance crews to address damaged roads and traffic signals more quickly. Self-driving cars could also potentially gather road safety data to help with the planning of improved traffic patterns and new road structures.

Some new advancements, such as Tesla’s HW2 Autopilot technology, allow self-driving semiautonomous vehicles to learn from human drivers. For example, a semiautonomous Tesla vehicle with this artificial intelligence system could essentially shadow a human driver’s actions and compare them to what the autonomous driving system would do in the same situation. When humans control self-driving cars in this manner, the AI programs following their actions can both refine their decision-making processes and identify errors in the driver’s judgment that may prevent safe vehicle operation and assume control to prevent an accident.

Are Self-Driving Vehicles Really Safe?

A completely safe vehicle does not really exist; every motor vehicle can cause harm in different situations. However, self-driving vehicles with refined autonomous processes are undeniably safer than human drivers. They can:

• apply shared learning between different individual vehicles
• update their algorithms and operations quickly and seamlessly with the introduction of new data
• learn from both the actions of human drivers
• learn from other self-driving vehicles that share their same software

Ultimately, the safety of a self-driving vehicle hinges on the software running the autonomous driving systems. Software, artificial intelligence, and machine learning systems are evolving very rapidly and the major manufacturers producing autonomous vehicles are investing tremendous amounts of industry capital into completely removing human error from U.S. roads. In time, self-driving vehicles will likely form the bulk of new vehicle sales and dramatically reduce the number of serious and fatal accidents in the U.S.