How Our Members Are Approaching Self-Driving Safety and Engineering

Self-driving vehicles are one of the latest milestones in a long history of safety technologies developed by the automobile and transportation industries. From hydraulic brakes and seatbelts to airbags, blind spot detection and backup cameras, the industry has made significant strides in making cars even safer.

Despite this tremendous progress, motor vehicle fatalities and injuries remain a major problem in the U.S. An estimated 36,120 people died in motor vehicle crashes last year, which is fortunately a slight decrease from previous years. Motor vehicle crashes are one of the leading causes of death for children and teens. One person dies every 50 minutes from drunk driving. More than 2,800 lives are lost to distracted driving. 

That’s why the full spectrum of the self-driving industry -- automakers, technology and robotics platforms, transportation networks -- is investing manpower, cutting-edge technology, artificial intelligence and billions of dollars into developing autonomous vehicles.

Experts agree that more than 90% of crashes are caused by human error: drunk driving, distracted driving, fatigue, speeding, and other decision and performance mistakes. By removing humans from the driving process entirely, fully autonomous vehicles are a significant safety technology that can help save lives and prevent injuries. Motor vehicle crashes are also an economic problem, resulting in an estimated $871 billion in economic loss and societal harm due to loss of life, pain, decreased quality of life, property damage, workplace losses and more.

The self-driving industry is thinking comprehensively and holistically about how best to develop and test this technology. Every simulation, every test on a public road, every rider survey helps the technology to get better and better.

Unfortunately, a recent study failed to take these kinds of conscientious engineering efforts into account. The study doesn’t include the fact that self-driving cars don’t speed or that the technology is building toward an ability to predict errors and perform evasive maneuvers better than humans. Based on its findings, the study actually demonstrates that self-driving vehicles could prevent 72% of crashes.

Some specific examples of how our Members are developing self-driving cars to be safe and superior to human driving include:

• Aurora is “proactively and intentionally taking the time to program our vehicles to operate like model drivers” when it comes to safety and following the rules of the road.” Aurora and other companies design self-driving cars to be “good citizens of the road” and avoid typical “jerk driver” behavior, such as

• avoiding the left lane of a highway

• avoiding passing vehicles on the right

• respecting the “every-other-rule” when merging

• avoiding blocking an intersection, and 

• making safer unprotected left turns.

• Lyft programs its self-driving technology’s decision making based on a “hierarchy of needs” with safety, legality and perceived safety prioritized far before comfort and route efficiency

• Nuro’s zero-occupant delivery vehicles reduce the number of people on the road, providing more protection from potential collisions or incidents while still getting needed supplies and groceries to homes and businesses

• Ford uses simulation to explore how to best communicate with passengers about what’s happening at any point during an autonomous ride to ensure passengers feel comfortable and confident.

• Waymo has driven 20 million miles autonomously on public roads across 25 U.S. cities and conducted over 15 billion miles in simulation in order to build an experienced safe, and  defensive driver Staying out of other drivers’ blind spots and giving extra room to the unpredictable movements of cyclists and pedestrians are among a few of their vehicles’ safe behavior decisions.

• The software on the Volvo cars that Uber uses for its self-driving testing is designed to continuously detect and track objects. For instance, another car’s left turn signal might be flashing yet the vehicle doesn’t actually turn left.

• While a human driver can only consider a few things at once using mirrors and looking out front, Argo AI explained how self-driving cars have a 360-degree view at all times, processing and predicting thousands of objects at a time

The industry is building the technology for a world where there are some autonomous vehicles, some human-driven vehicles, along with all the unplanned and unpredictable challenges of bicyclists, pedestrians, weather, environment and more.

We agree with the study authors that self-driving vehicles must be designed with safety first and foremost, and that is precisely what our members are developing and testing. Self-driving vehicles are another crucial step toward saving lives and making our roads safer for all.