Robotaxis crash less often: why researchers still hold back on the verdict

About 50 million driverless miles have given Waymo a clear edge, though not a universal victory for autonomous technology. An IIHS study found that the company’s robotaxis were involved in 68% fewer crashes of the kind an ordinary driver would most likely report to the police. Injury crashes were down 81%, single-vehicle crashes down 85%.

Researchers compared Waymo’s operations in Phoenix, San Francisco, Los Angeles and Austin with ordinary drivers in the same cities over the same window — from 2021 to 2024. The outcome depended heavily on location: crash rates were 76% lower in Phoenix, 71% lower in Los Angeles and 35% lower in San Francisco. In Austin the figure came out 4% higher than for humans, but the sample there is too small for conclusions.

The method matters more here than the headline percentage. Out of 736 reported incidents involving automated vehicles, IIHS treated only 22% as comparable to the crashes people normally report to the police. The robotaxi mileage was set against roughly 222 billion miles covered by humans in the same places. Crash frequency could be calculated for Waymo alone: the company discloses its mileage voluntarily, while federal rules do not require it from every operator.

The conclusion applies only to SAE Level 4, which drives the car on its own within mapped areas. It is not a verdict on the widespread Level 2 assistants, where a human must keep watching the road at all times. 32CARS.RU has previously explained how remote operators help Waymo vehicles in tricky situations: specialists offer guidance but never take the wheel in place of the automation.

Waymo has received independent confirmation of a lower crash rate inside its operating zones. Yet without mandatory mileage reporting and a single standard for logging incidents, the 68% figure remains proof for one operator rather than a safety certificate for the whole industry.