Tesla robotaxi service logs 14 crashes over roughly 800,000 miles
- Tesla’s robotaxi service had logged 14 reported crashes over about 800,000 miles as of August 6, 2026, according to Autonocion’s report. - Waymo said it serves more than 500,000 fully autonomous trips a week, while co-CEO Dmitri Dolgov called camera-only autonomy “weak sensing.” (autonocion.com) - NHTSA’s crash-reporting order remains in force, and Waymo’s safety dashboard continues publishing rider-only mileage totals. (nhtsa.gov)
Autonocion reported on August 6 that Tesla’s robotaxi service has logged 14 crashes across roughly 800,000 miles, with a paid human sitting in the front seat during service. The report framed the figure against Waymo’s larger U.S. robotaxi operation, which Waymo says now handles more than 500,000 fully autonomous trips a week across 10 U.S. cities. Waymo’s public safety dashboard says the company had driven 220.6 million rider-only miles through March 2026 without a human driver in the vehicle. (autonocion.com) The comparison lands in a debate that has split the industry for years: Tesla relies primarily on cameras, while Waymo uses cameras, lidar and radar together. (nhtsa.gov) Waymo co-CEO Dmitri Dolgov said this week that camera-only autonomy relies on what he called “weak sensing,” arguing that such systems hit a safety ceiling before reaching full autonomy. Electrek reported Dolgov made the remarks during a Y Combinator Startup School appearance. ### Where does the 14-crash figure come from? The 14-crash figure cited by Autonocion appears to track incidents reported under the National Highway Traffic Safety Administration’s Standing General Order, which requires identified manufacturers and operators to report certain crashes involving automated driving systems or Level 2 driver-assistance systems. (autonocion.com) NHTSA says the order is intended to give the agency timely notice of real-world crashes involving those systems. Several secondary reports this year, including Electrek and other industry outlets, said Tesla’s Austin robotaxi fleet had accumulated roughly 800,000 miles since launching in June 2025 and had reported 14 crashes to NHTSA by early 2026. (electrek.co) Those reports also said the service still used a paid human in the front seat, rather than operating rider-only. Tesla did not appear in the sourced material here with a fresh public statement on the August 6 comparison. ### How is Waymo describing its own scale? (nhtsa.gov) Waymo said in a March 31 company post that it was serving over half a million trips every week across 10 U.S. cities. A separate Waymo page says the company’s rider-only mileage reached 220.6 million miles through March 2026. Those figures describe fully autonomous service without a human driver in the vehicle, which is the benchmark Autonocion used in contrasting the two companies. Waymo has also used those mileage totals to publish comparative safety claims. (electrek.co) In a March 19 update, the company said an analysis covering more than 170 million fully autonomous miles showed 92% fewer crashes causing serious or fatal injuries than human drivers in the same operating domains. That claim is Waymo’s, not an independent federal ruling. ### What exactly did Dmitri Dolgov say about cameras? Electrek reported on August 4 that Dolgov argued camera-only self-driving “falls short” because “weak sensing” reaches a limit before it gets to what he described as superhuman safety. (waymo.com) Autonocion characterized the same argument more bluntly, saying Dolgov had made the case that a camera-only approach is a dead end. Waymo’s own technical materials describe its system as fusing camera, lidar and radar inputs. That criticism matters because Tesla Chief Executive Elon Musk has long argued that cameras are sufficient for autonomy and has rejected lidar as unnecessary. (waymo.com) NHTSA is also investigating Tesla’s handling of reduced-visibility conditions in vehicles using its camera-based approach, according to agency documents. ### Why is the comparison hard to make cleanly? NHTSA’s Standing General Order data are disclosure records, not a single standardized safety scorecard, and the companies operate different systems in different conditions. (electrek.co) Tesla’s reported figure in this comparison covers a much smaller mileage base and includes service with a paid human in the front seat, while Waymo’s published totals refer to rider-only operations at much larger scale. Waymo’s next public updates are likely to appear on its safety dashboard and company blog, while Tesla crash disclosures covered by the federal reporting order would continue to flow through NHTSA’s reporting system. (static.nhtsa.gov) (waymo.com) (nhtsa.gov)