# Comparison of Waymo Rider-Only Crash Data to Human Benchmarks at 7.1 Million Miles

arXiv: https://arxiv.org/abs/2312.12675 · HTML: https://mlsanity.com/paper/2312.12675/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 20 Dec 2023.

## Built on by (most cited first, 8 of 8)

- [CtRL-Sim: Reactive and Controllable Driving Agents with Offline Reinforcement Learning](https://mlsanity.com/paper/2403.19918/) (2024)
- [CurricuVLM: Towards Safe Autonomous Driving via Personalized Safety-Critical Curriculum Learning with Vision-Language Models](https://mlsanity.com/paper/2502.15119/) (2025)
- [CRASH: Challenging Reinforcement-Learning Based Adversarial Scenarios For Safety Hardening](https://mlsanity.com/paper/2411.16996/) (2024)
- [Navigating the Edge with the State-of-the-Art Insights into Corner Case Identification and Generation for Enhanced Autonomous Vehicle Safety](https://mlsanity.com/paper/2503.00077/) (2025)
- [Credibility-Weighted Pricing of Autonomous Vehicle Liability Under Operational Design Domain Shift](https://mlsanity.com/paper/2606.17451/) (2026)
- [OmniSCS: Omni Safety-Critical Scenario Synthesis for Autonomous Driving via a Fully Editable Driving World](https://mlsanity.com/paper/2607.09764/) (2026)
- [When2Talk: When Should a Proactive In-Car Agent Talk?](https://mlsanity.com/paper/2609.12503/) (2026)
- [When Compliance Data Masquerades as Evaluation: Measurement Validity for Deployed AI Systems](https://mlsanity.com/paper/2609.13642/) (2026)

## Abstract (arXiv)

This paper examines the safety performance of the Waymo Driver, an SAE level 4 automated driving system (ADS) used in a rider-only (RO) ride-hailing application without a human driver, either in the vehicle or remotely. ADS crash data was derived from NHTSA's Standing General Order (SGO) reporting over 7.14 million RO miles through the end of October 2023 in Phoenix, AZ, San Francisco, CA, and Los Angeles, CA. When considering all locations together, the any-injury-reported crashed vehicle rate was 0.6 incidents per million miles (IPMM) for the ADS vs 2.80 IPMM for the human benchmark, an 80% reduction or a human crash rate that is 5 times higher than the ADS rate. Police-reported crashed vehicle rates for all locations together were 2.1 IPMM for the ADS vs. 4.68 IPMM for the human benchmark, a 55% reduction or a human crash rate that was 2.2 times higher than the ADS rate. Police-reported and any-injury-reported crashed vehicle rate reductions for the ADS were statistically significant when compared in San Francisco and Phoenix, as well as combined across all locations (except for any-injury-reported in Phoenix). The any property damage or injury comparison had statistically significant decrease in 3 comparisons, but also non-significant results in 3 other benchmarks. Given imprecision in the benchmark estimate and multiple potential sources of underreporting biasing the benchmarks, caution should be taken when interpreting the results of the any property damage or injury comparison. Together, these crash-rate results should be interpreted as a directional and continuous confidence growth indicator, together with other methodologies, in a safety case approach.