🤖 AI Summary
Waymo published safety data covering 96 million miles through June and NHTSA crash reports through Aug. 15 show that of the 45 most serious Waymo crashes between mid‑February and mid‑August, the large majority were not caused by Waymo’s self‑driving software. Waymo reported 34 airbag‑triggering crashes versus an estimated 159 for typical human drivers in the same areas (a 79% reduction); injury‑causing crashes were ~80% lower, pedestrian injuries 92% lower, and cyclist injuries 78% lower. The most clearly Waymo‑linked incident was a mechanical failure (a front wheel detached), not a perception or planning error.
A closer read of the 45 NHTSA reports finds 41 crashes involving driving errors but 37 of those appear mostly or completely the fault of other road users: 24 collisions occurred while Waymos were stationary (19 rear‑ends), seven rear‑ended a moving Waymo, and several involved other drivers cutting into Waymo’s right of way or sideswiping in narrow streets. Three “dooring” injuries and one ambiguous multi‑car chain reaction (after a Waymo braked for a cat) highlight non‑software risks—passenger behavior, mechanical reliability, narrow urban geometry and small‑object detection. The data bolster Waymo’s safety claims but also underscore the limits of AVs: operator transparency and independent scrutiny remain essential, and improvements in vehicle robustness, exit‑warning UX, and edge‑case perception could further reduce residual harms.
Loading comments...
login to comment
loading comments...
no comments yet