A first-of-its-kind study has revealed that autonomous vehicle technology reduced bodily injury claims frequency by 100 percent and property damage claims frequency by 76 percent.

Global reinsurer Swiss Re and automated driving company Waymo partnered on the study.

The two compared Waymo’s third-party liability claims data with mileage and ZIP-code-calibrated, human-driven private passenger vehicle baselines established by the insurer for the period of 2016-2021, from over 600,000 claims and over 125 billion miles of exposure.

The study indicates the use of liability insurance claims provides a more comprehensive assessment than collision databases from police reports because claims data has more consistent standards for reporting along with a higher frequency of reporting, as well as information relating to crash and injury causation. In addition, police reports don’t always capture non-collision related injuries.

The study found that the Waymo Driver significantly improved road safety over 3.8 million fully autonomous miles driven in San Francisco and Metro Phoenix.

The difference between the autonomous functions is as follows:

  • In “Manual” mode, the Waymo vehicle is driven completely manually (i.e., without the ADS engaged) by an autonomous specialist (human driver) who is capable of reacting to a dynamic environment and operating vehicles under strict safety guidelines.
  • In testing operations “TO” mode, the ADS is engaged to operate the vehicle under monitoring of trained human autonomous specialists.
  • A collision that follows after the autonomous specialist takes over control of the driving task does not count as a Manual claim but as a TO claim if the ADS performed a driving maneuver that placed the vehicle in the situation that led to the collision, the study noted.
  • A collision is categorized under TO as long as the ADS was engaged at any time during the five seconds leading up to the impact. A collision could still be categorized under TO under the 5-second rule even if a human maneuver may have led to the collision, leading to a conservative estimate.

When Waymo vehicles were driven manually by trained human drivers for 14,436,298 miles for data collection, bodily injury claims frequency was reduced by 45 percent compared to the Swiss Re human driver baseline, while property damage claims frequency was significantly reduced by 34 percent (2.22 vs 3.34 claims per million miles).

While driving 35,228,320 miles in TO mode, the Waymo Driver, together with autonomous specialists, significantly reduced bodily injury claims frequency by 92 percent.

While driving without a human behind the steering wheel in rider only (RO) mode for 3,868,506 miles, the Waymo Driver reduced bodily injury claims frequency by 100 percent, while property damage claims frequency was significantly reduced by 76 percent.

When TO and RO datasets were combined, totaling 39,096,826 miles, there was a significant reduction in both bodily injury and property damage claims frequency, by 93 percent.

The researchers concluded there is strong evidence proving the ability of ADS to reduce bodily injuries on public roads.

The purpose of the joint study was to construct performance baselines to benchmark autonomous vehicle safety performance. Researchers indicate the study method used overcomes existing challenges facing autonomous and human driver crash rate comparisons and can be applied within the industry to assess the safety performance of future autonomous vehicle deployments.