An Analysis of Traffic Collisions Involving Autonomous (“Self-Driving”) Vehicles in San Francisco, California.

by Michael Bonilla

This study performed an analysis of the cumulative collision data for Autonomous Vehicles that are currently approved Deployment Permit Holders in California (Cruise LLC and Waymo, LLC). The analysis compared the publicly available safety data of relative conventionally driven vehicles in San Francisco, California and the United States (adjusted for underreporting of collision that do not involve fatality). My research hypothesis proposes that if human driving is augmented by artificial intelligence, it can eliminate the vast majority of automobile vehicle accidents. This, in turn, will most likely lead to the reduction or removal of “Third-Party” coverages, such as Property Damage Liability and Bodily Injury Liability.
From my research, it showed that Autonomous Vehicles were involved in crashes at a significantly lower (-71%) property damage only crash rate compared to the San Francisco Municipal Transportation Authority fleet. Autonomous Vehicles (AV) were involved in crashes at a significantly higher rate than both the California (+968%) and National (+524%) Conventional Vehicles (CV) fleets. Autonomous Vehicles had a higher injury crash rate per 1,000,000 vehicle miles traveled. Autonomous Vehicles outperformed Conventional Vehicle fleets by demonstrating a Fatality crash rate of 0 per 1,000,000 Vehicle Miles Traveled (VMT). During the period of time we evaluated, the vast majority (94.9%) of Autonomous Mode accidents resulted in the Conventional Vehicle being the most likely At-Fault party or no reasonable likelihood of contributable negligence on behalf of the Autonomous Mode vehicle towards the crash.

57 pages, Kindle Edition

Published April 14, 2023