The advent of fully autonomous vehicles and autonomous vehicle technology has long been hailed by proponents as a means towards preventing, or even completely eliminating, automobile crashes. In fact, General Motors has pointed to autonomous tech as a means towards achieving the “Zero Crashes” portion of its Zero Crashes, Zero Emissions, Zero Congestion commitment. However, the ability for autonomous vehicles to prevent crashes depends largely on their programming, as outlined by a new study from the Insurance Institute for Highway Safety (IIHS).
The new IIHS study looked at more than 5,000 police-reported crashes from the National Motor Vehicle Crash Causation Survey, as collected by the National Highway Traffic Administration (NHTSA). The IIHS team identified five primary contributors to a crash, including:
- Sensing and perceiving, including failure to recognize a hazard
- Predicting, such as driver misjudgment of another vehicle’s speed
- Planning and deciding, such as driving too fast for the conditions
- Execution and performance, such as mistakes in control of the vehicle
- Incapacitation, including impairment as a result of the consumption of drugs or alcohol
According to the study, autonomous vehicles would only be able to avoid crashes resulting from two of these categories, namely sensing and perceiving, and incapacitation, given the current state of AV programming.
“Crashes due to only sensing and perceiving errors accounted for 24 percent of the total, and incapacitation accounted for 10 percent,” states the IIHS. “Those crashes might be avoided if all vehicles on the road were self-driving – though it would require sensors that worked perfectly and systems that never malfunctioned. The remaining two-third might still occur unless autonomous vehicles are also specifically programmed to avoid other types of predicting, decision-making and performance errors.”
The IIHS suggests that autonomous vehicles be programmed specifically to prioritize safety over speed and convenience, rather than to make the same decision about risk as human drivers.
As the organization points out, this may put the preference of the rider or riders at odds with the safety priorities of the autonomous vehicle system, but is critical in order for AVs to live up to their safety promises.
Back in January, General Motors debuted Cruise Origin, its first fully driverless car. Origin has been in development for over three years, and is designed from the ground-up as a completely autonomous all-electric vehicle capable of transporting up to four passengers. By the time it goes into production, GM says Cruise Origin will have core software that will best an average human driver in terms of performance and safety.