Autonomous rideshare operator Waymo released safety data showing its vehicles achieved an 82 percent reduction in injury-causing crashes across 271 million driverless miles in San Francisco, Los Angeles, Austin, Atlanta, and Phoenix, compared to human driver benchmarks.
How Waymo Evaluates 271 Million Driverless Miles
Operating across San Francisco, Los Angeles, Austin, Atlanta, and Phoenix, the Google-funded startup utilized its accumulated data to evaluate the performance of the Waymo Driver against human-driven baselines.
The safety evaluation compares Rider-Only crash rates directly to human crash benchmarks for surface streets over identical distances to equalize per-mile comparisons. According to the company, the automated driving system was involved in 841 fewer injury-causing crashes than would be expected from flesh-and-blood drivers operating in the same regions.
Detailed Crash Metrics Across Vulnerable Road Users
The safety impact measured by the company’s research highlights significant reductions in incidents involving vulnerable road users. Compared to human crash data, the automated driving system reduced injury-causing crashes involving pedestrians by 93 percent, cyclists by 86 percent, and motorcyclists by 82 percent.
- Pedestrian Incidents: 93 percent reduction in injury-causing crashes.
- Cyclist Incidents: 86 percent reduction in injury-causing crashes.
- Motorcyclist Incidents: 82 percent reduction in injury-causing crashes.
- Serious Injuries: 95 percent reduction in crashes causing serious injury or worse.
- Airbag Deployments: 82 percent fewer airbag deployments in any vehicle crashes.
Regional variations also emerged in the findings. Compared to human crash data, Waymo vehicles recorded 8 percent fewer crashes resulting in airbag deployments in Atlanta and 10 percent fewer crashes resulting in injury in San Francisco.
Methodological Differences Between Autonomous and Human Reporting
A central challenge in evaluating autonomous vehicle safety involves accounting for structural differences in how crashes are documented. Autonomous operators like Waymo must report any physical contact resulting in property damage. In contrast, human crash reporting typically depends on a specific damage threshold or a police report to enter official databases.
To establish a consistent baseline, Waymo utilized police report databases as the primary human crash data source, pairing them with Vehicle Miles Traveled statistics in its operating regions. The National Highway Traffic Safety Administration estimates that roughly 60 percent of property damage crashes and 32 percent of injury crashes go unreported to authorities entirely, making standardized comparisons complex.
The Insurance Institute for Highway Safety praised the autonomous operator for its open-data policy. By making detailed information regarding crashes and miles driven publicly accessible, the company aims to support independent research and foster public trust.
The Path Forward for Autonomous Rideshare Safety
While the debate surrounding the comparative safety of autonomous systems versus human drivers continues, the expanding mileage dataset provides quantitative metrics for ongoing evaluation. Waymo acknowledged that continuous refinement remains necessary and emphasized that autonomous deployment is only one component of addressing broader traffic safety challenges.
