Comparative Safety Performance of Autonomous- and Human Drivers: A Real-World Case Study of the Waymo One Service (2309.01206v1)

Abstract: This study compares the safety of autonomous- and human drivers. It finds that the Waymo One autonomous service is significantly safer towards other road users than human drivers are, as measured via collision causation. The result is determined by comparing Waymo's third party liability insurance claims data with mileage- and zip-code-calibrated Swiss Re (human driver) private passenger vehicle baselines. A liability claim is a request for compensation when someone is responsible for damage to property or injury to another person, typically following a collision. Liability claims reporting and their development is designed using insurance industry best practices to assess crash causation contribution and predict future crash contributions. In over 3.8 million miles driven without a human being behind the steering wheel in rider-only (RO) mode, the Waymo Driver incurred zero bodily injury claims in comparison with the human driver baseline of 1.11 claims per million miles (cpmm). The Waymo Driver also significantly reduced property damage claims to 0.78 cpmm in comparison with the human driver baseline of 3.26 cpmm. Similarly, in a more statistically robust dataset of over 35 million miles during autonomous testing operations (TO), the Waymo Driver, together with a human autonomous specialist behind the steering wheel monitoring the automation, also significantly reduced both bodily injury and property damage cpmm compared to the human driver baselines.

• The paper demonstrates that Waymo One’s autonomous service significantly reduces liability claims compared to human-driven vehicles over millions of miles.
• It employs standardized insurance claims data to compare bodily injury and property damage claims, revealing up to 93% reduction in accidents.
• Implications include potential for lower insurance premiums and regulatory shifts promoting wider adoption of autonomous driving technology.

Comparative Safety Performance of Autonomous- and Human Drivers: A Real-World Case Study of the Waymo One Service

This paper presents a detailed examination of the safety performance of the Waymo One autonomous driving service in comparison to human drivers, utilizing third-party liability insurance claims as a metric. The paper methodically contrasts the incidence of bodily injury and property damage claims between autonomous and human-driven vehicles, leveraging a robust dataset from Swiss Reinsurance Company and Waymo LLC.

Methodology and Data

The analysis is grounded in a significant dataset comprising over 3.8 million miles driven autonomously in rider-only (RO) mode and over 35 million miles in testing operations (TO) where a human specialist supervises the ADS. For comparison, a human driver baseline is developed using Swiss Re's insurance claims data, normalized to account for mileage and geographical variances, specifically targeting operating areas in San Francisco and Phoenix.

The paper outlines the methodology to create standardized comparisons, focusing on the comprehensive nature of liability insurance claims as opposed to police reports, which often miss non-collision-related injuries. Claims data provide insights into crash causation through structured adjudication processes, allowing for a robust assessment of safety performance.

Key Findings

1. Rider-Only Mode (RO): The Waymo Driver demonstrated an impressive safety performance, with zero bodily injury claims across 3.8 million miles, compared to a human driver baseline of 1.11 claims per million miles. Property damage claims were also reduced to 0.78 cpmm, a 76% reduction from the human baseline of 3.26 cpmm.
2. Testing Operations Mode (TO): The dataset evidenced a 92% reduction in bodily injury claims (0.09 cpmm vs. 1.09 cpmm baseline) and a 95% reduction in property damage claims (0.17 cpmm vs. 3.17 cpmm baseline).
3. Combined TO and RO Modes: Across 39 million miles, there was a 93% reduction in bodily injury claims (0.08 cpmm vs. 1.09 cpmm baseline) and property damage claims (0.23 cpmm vs. 3.17 cpmm baseline).
4. Manual Operations: Even when driven manually by trained specialists, Waymo vehicles showed a reduction in bodily injury claims by 45% and property damage by 34%, although only the latter was statistically significant.

Implications

The results provide compelling evidence of the enhanced safety performance of autonomous vehicles, specifically the Waymo One service, over conventional human driving. This has significant implications for the broader adoption of autonomous driving technology, potentially influencing regulatory frameworks and insurance industry standards. The reduction in claims could result in lower insurance premiums and a reevaluation of risk assessments associated with autonomous vehicles.

Future Research Directions

Further research could explore in-depth analyses tailored to different operational design domains (ODDs) to refine human driving baselines. While this paper limits freeway data, understanding autonomous vehicle performance across diverse roadway environments remains crucial. Additionally, longitudinal studies may capture evolving trends as autonomous technologies continue to develop.

In conclusion, this comprehensive paper substantiates the safety advantages of Waymo’s autonomous driving service, providing a template for future evaluations of autonomous vehicle safety using insurance data as a benchmark for performance measures.