Waymo's autonomous vehicles have now logged over 170 million miles on public roads, a distance the company equates to 200 human lifetimes of driving. The central claim from its latest data update is stark: its vehicles are involved in 92% fewer crashes resulting in serious injury or death compared to human drivers. With a fleet of roughly 3,000 vehicles across ten cities, Waymo calculates its technology now prevents a serious-injury crash about once every eight days.

Yet, safety advocates argue this impressive statistic presents an incomplete picture. A recent analysis by the Advocates for Highway and Auto Safety highlights contextual gaps. They note that 45% of the crashes Waymo reports involve no passengers, which automatically lowers potential injury rates. Furthermore, about 80% of its incidents are rear-end collisions, a rate far above the national average for human drivers.

"Claiming a safety benefit from crashes where no occupant is present is incongruous if the goal is to transport people," said Cathy Chase, the group's president, in a letter to lawmakers.

Real-world incidents underscore the debate. Federal investigators are examining cases where Waymo vehicles drove past stopped school buses and one where a robotaxi, traveling at low speed, struck a child in Santa Monica, resulting in minor injuries. Other reports, like a vehicle briefly blocking an ambulance during a crisis in Austin, often fall outside official crash reporting mandates.

Waymo maintains its methodology is sound, considering injuries to anyone involved in a crash sequence, including pedestrians. The company has submitted its data for peer-reviewed journal publication, seeking external validation.

As the fleet's mileage grows, so does the scrutiny. The fundamental question remains whether Waymo's numbers, while compelling, fully capture the complex safety profile of a technology still learning to navigate a world built for humans.

Source: The Verge