According to the U.S. National Safety Council, in 2018, 40,000 people died in car crashes, a 1 percent decline from 40,231 deaths in 2017 and 40,327 deaths in 2016. An estimated 4.5 million people were seriously injured in car crashes in 2018, also a 1 percent decrease from 2017 figures. These casualties are a public health issue, and impose billions of dollars in private and social costs. Autonomous driving technology has the potential to substantially reduce this human toll.
Many factors contributed to reducing the rate of crashes, injuries, and fatalities—including the gradual adoption of on-vehicle safety technologies. These systems were introduced in various model years: modern frontal airbags in 1984, antilock brakes in 1985, electronic stability control in 1995, head-protecting side airbags in 1998, and forward-collision warnings in 2000. But it typically takes three decades for safety features that start out on luxury vehicles to reach the entire vehicle fleet, as older vehicles are replaced with newer models.
The U.S. Insurance Institute for Highway Safety (IIHS) estimated that if all vehicles had forward collision and lane departure warning, sideview (blind spot) assist, and adaptive headlights, nearly a third of crashes and fatalities could be prevented. These features are generally associated with Level 0 or Level 1 vehicle automation.
Dynamic brake support, a Level 1 feature, reduces stopping distances for drivers who have made a decision to stop quickly. This will improve safety outcomes, but will not cure driver error in situations where no decision is made to stop. As vehicle automation technology advances from function-specific automation (Level 1) to Combined Function Automation (Level 2), the driver can cede active primary control in some situations, and at least two functions can be automated to work in unison.
Level 3 automation allows the driver to cede full control of all safety-critical functions in certain situations. Crashes, injuries, and fatalities due to driver error under this condition would likely be substantially reduced. Level 3 vehicles also might drastically reduce the number of crashes, injuries, and fatalities involving motorcycles, pedestrians, and cyclists, as vehicles automated at this level will not be distracted, impaired, or reckless—and can increase avoidance of others who might be. However, Level 2 and 3 automation also might increase some categories of crashes if consumers rely upon the systems too much.
There may be further risk reductions with the transition from Level 3 to Level 4 and Level 5automation, especially in the number of alcohol-related crashes. For example, alcohol was involved in more than 39 percent of motorist fatalities, and roadway safety could improve exponentially when these impaired drivers cede control to fully self-driving auto-mated vehicles. Eliminating up to a third of traffic deaths through vehicle automation just by limiting alcohol-impaired drivers would represent a dramatic improvement in roadway safety.
In short, autonomous driving technology will likely lead to substantial reductions in crashes and the resulting human toll. While a portion of these benefits will accrue to the purchaser of the vehicle, much of the benefit is in the form of a positive externality to other vehicles, pedestrians, and bicyclists.