“In its first report on its autonomous vehicle operations in Phoenix, Arizona, Waymo said that it was involved in 18 crashes and 29 near-miss collisions…” reports the Verge
The 18 refer to actual crashes where actual contact was made was contacted with another vehicle (or potentially if it has been, or bicyclist). There were no injuries.
The 29 near-misses refer to simulated responses of their AV system; which didn’t actually happen, because the human driver intervened. The vast majority of Miles logged were with a human “safety driver” present.
By the way here’s the story of the kook that caused one of the crashes, he was apparently a disgruntled ex-Waymo driver ; it also includes a link to the nut who brandished a gun at a waymo driver, apparently because he was aggreived by the waymo vehicles.
Replaying real life: how the Waymo Driver avoids fatal human crashes (#fatalsim)
In March of 2021 Waymo released an interesting simulation of how their vehicle would behave in each the 72 fatal vehicle collisions occurring in their geo-fenced area (primarily part of the City of Chandler) over a 12 year period. Here’s a news item from Verge the claim or hope being “The Google spinoff discovered that replacing either vehicle in a two-car crash with its robot-guided minivans would nearly eliminate all deaths”.
There was a longer news item in the Arizona Republic; that piece is littered (16 occurrences!) with the ‘a-word’ — crash not accident people, including at least one flagrant DUI
…when negligence is claimed or proven, avoid accident, which can be read by some as a term exonerating the person responsible — AP Stylebook
Both the Verge news item, the waymo post, and the waymo scholarly article all manage to avoid the offensive term