Self-Driving Cars: Are They Actually Safe?

Cars with beams to illustrate their smart sensing technology
Public opinion on the safety of self-driving cars dipped in 2018, when Elaine Herzberg, a pedestrian crossing the street was struck and killed by a partially-automated vehicle. The collision was scrutinized, and a variety of elements were discovered that directly influenced the impact.

There were two control systems involved that were responsible for the crash: the computer system and the human safety driver.

It was discovered that the computer system itself did, in fact, detect the pedestrian, but also that the software determined no immediate reaction was necessary. Additionally, the emergency driver that was involved had been warned by the car’s auto pilot system multiple times to return their hands to the wheel by the program.

As you can see in the above video footage, the driver was looking down and not at the road.

Neither the automated vehicle nor the driver slowed down or took measures to avoid the incident.

If an algorithm can detect a potential accident but may miscalculate distance, and a human is capable of avoiding accidents but may not be paying attention, which is more safe — or dangerous — to rely on for safety?

And, regardless of your vehicle’s self-driving capabilities, you – as the driver – are 100% responsible for purchasing and maintaining car insurance to protect you (and your financial stability) if you are involved in an accident. Compare a few of the best policies in your area by entering your zip code into our free quote comparison tool.

Self-Driving Cars Safety Statistics

When considering self-driving car safety statistics, it is important to recognize the five levels of automation that describe the amount of human or automated assistance at play in a vehicle.

There are currently no fully-automated vehicles that can drive without assistance in every circumstance.

Even when the self-driving vehicle is controlling the steering, braking, accelerating, and navigation, an emergency driver or human assistance is still important to the operation of the vehicle and its ability to function safely.

Statistics dedicated to the safety of self-driving cars in relation to conventional vehicles are difficult to collect and analyze when considering the lack of historical data surrounding self-driving cars. Another complication arises when determining how to assess and discover fault in an accident involving automated vehicles.

 – Determining Who (or What) is at Fault

It is important to distinguish between fully-automated self-driving car accidents, accidents that occur with a self-driving car where the automation is not at fault, or when the fault lies with the other driver and conventional car.

The National Highway Traffic Safety Administration (NHTSA) and the Bureau of Transportation Statistics state that 94% of serious crashes that occur are due to human error.

37,133 people died in conventional vehicle-related rashes in the United States in 2017.

NHTSA also estimates that 30% of conventional motor vehicle accidents are rear-end fender benders involving distracted drivers.

A data analysis, funded by the San Jose State University College of Engineering, and using data presented by the NHTSA and the Federal Highway Administration (FHWA), found 62% of self-driving vehicle accidents were rear-end collisions.

This 50% increase in rear-end accidents from conventional to self-driving cars may look dramatic, but when the data was further analyzed, only on one occasion was the automated vehicle responsible for another vehicle.

The same analysis showed the total number of accidents for self-driving vehicles in 2015 occurred at a rate of 26 in 1,088,453 miles, while conventional vehicle total accidents occurred 6,296,000 times out of every 3.148 trillion miles traveled. Breaking down this data would show the following:

Self-driving cars would average an accident every 42,017 miles, while conventional vehicles would only average one accident every 500,000 miles.

However, the data may be misleading when considering that, in 22 out of the 26 reported accidents involving autonomous vehicles, the self-driving car was not at-fault; either the driving automation had been disengaged prior to the collision, or a conventional vehicle was at fault.

Wired performed a review of the 104 autonomous vehicle collision reports the state of California has received from developers since 2014, 49 of which are from 2018. Though most incidents are minor, it is important to recognize the 2018 increase correlates with the expansion of self-driving fleets making it onto the streets, and the allowance of self-driving cars into public spaces rather than only in controlled environments.

The data review further shows that 28 of the 49 filed reports were cases of self-driving cars being rear-ended. In 22 out of those 28 incidents, the automated vehicle was in control. Of note, California law states that the fault lies with the rear vehicle in the instance of a fender bender.

– Technology vs Human Behavior

It is important to notice the behavior of self-driving cars that may not be coinciding with human driving etiquette.

Self-driving cars may make quick, jerking motions or other reactionary and precautionary sudden stops, as opposed to changing lanes or other more human evasive tactics and behaviors.

This difference in reaction to traffic or other variables may be one of the main contributors to the reported collisions.

Human behavior, though it may be reactive in certain situations (assessing a car stopping in front of them and choosing to change lanes rather than quickly decelerate or stop), can also bring about dangerous conditions (aggressive driving, road rage, or DUI). Consider a few statistics from The Ultimate List of Driving Statistics for 2019 collected by Driving-Tests.org:

  • Driver-related factors are present in 90% of crashes. This includes distraction, fatigue or other impairments, and driver performance error.
  • Aggressive driving causes 66% of traffic fatalities.
  • Tailgating contributes to more than one-third of all crashes.
  • Teen drivers do not consider speeding 5-10 miles over the limit as dangerous.
  • Crash risk is highest during the first year of licensed driving.

Teens and other first-year license holders have the highest risk of crashing due to inexperience, and are often at the mercy of higher insurance premiums. On the other end of the spectrum, elderly drivers often receive much lower premiums after a lifetime of driving experience, but must also assess the safety of their driving abilities.

The American Driving Survey from AAA determined the following from 2016-2017:

American drivers were driving an average of 51 minutes per day, for an average of 31.5 miles.

Both the time spent and miles driven saw a near 5% increase from 2014-2015.

The same survey found that the “driving-aged” population has increased every year from 2014-2017 to a total of 3.6 million. With the increasing distance, time, and the number of American drivers on the road, any particular driver’s life stage, abilities, and record of previous errors are increasingly important in terms of safety, and will play heavily in the driver’s ability to be insured.

Pros and Cons of Self-Driving Cars

Fully self-driving cars, where the vehicle can drive and accommodate all circumstances, have yet to be realized.

The current status of self-driving cars is one that utilizes technology, yet still relies on human instinct and oversight during emergency situations.

As technology advances, so will the conversation surrounding concerns and risks.

Dangers of Self-Driving Cars

  • Electronic malfunctions, faulty sensor controlling, and potential miscalculations in software
  • Cybersecurity vulnerability to hacking and remote control
  • Unregulated or lack of strict safety standards due to the emerging unknowns in the industry
  • Exposure to an abundance of electromagnetic field radiation from myriad power accessories like GPS, remote controls, Bluetooth, Wi-Fi, and other components
  • A false sense of security by drivers who reduce attention on the road and rely too heavily on automation

Benefits of Self-Driving Cars

  • Reduction of driver error, which factors into 94% of crashes
  • Reduction of traffic fatalities caused by aggressive drivers
  • Independence for senior citizens no longer capable of safe driving
  • Independence for self-sufficient people with disabilities that inhibit their ability to drive
  • Increased transportation affordability through ride-sharing
  • Increased transportation affordability for drivers with expensive insurance premiums
  • Reduced traffic congestion due to digital communication between vehicles and sophisticated collision-avoidance systems
  • Environmental gains such as lowered greenhouse gases and emissions via the elimination of needless idling, reduction of cars in use through ride-sharing, and heightened fuel efficiency

The Future of Self-Driving Cars

The future of self-driving cars is unknown. How will they be insured, and what type of insurance should they be covered under? Who will be liable in the event of an accident? Is self-driving software adaptable enough to accommodate human drivers and reactions on the road?

With nearly 40,000 deaths per year, and only 6% of those deaths not relating to human error, the push for self-driving cars and reliance on automation continues to swing on the pendulum of public opinion.

Even with the loss of public trust due to accidents involving self-driving vehicles, the potential for this new technology to save lives is unprecedented. The path toward 100% safe autonomous vehicles may be rocky, but the destination — a time when nobody has to feel unsafe on or near roads, whether driver, passenger, or pedestrian — will be worth the journey.

Remember, no human or driverless technology is perfect. Car insurance will always be essential, and – in all states but two – it’s required by law. Get started now in find the best coverage for you – our free tool will simplify the process for you.