The Paul Douglas Institute

Trends in Collision and Disengagement Data

This report compares California’s regulations for ADS testing with a safety driver to five other states’ regulations. It analyzes self-reported disengagement and mileage totals for ADS equipped vehicles testing in California between 2014 and 2018 and collisions involving ADS equipped vehicles which occurred on or before April 14, 2019. Rear-end collisions made up 72% of reported collisions in autonomous mode. Approximately 78% of collisions resulted in minor damage. All reported collisions occurred within the Bay Area. Most Cruise collisions occurred in downtown San Francisco and most Waymo collisions occurred around Mountain View. Approximately 97% of collisions occurred on an urban street. Large numbers of a sample of 2018 disengagements listed issues with the ADS technology in the disengagement cause description. Software, planning, and perception discrepancy were each reported in disengagement causes significantly more than other words or phrases. Most of the disengagement cause descriptions which mentioned an object did not describe what the object was. External factors such as traffic lights or lane markings occurred much less frequently. There are significant differences in the requirements that companies must meet to test ADS with a safety driver from state to state. While the permitting process in California is similar to other states, test driver certification is generally more comprehensive. Of the six states compared in this report, disengagement reporting is unique to California, and only California and Pennsylvania require that entities simulate road testing conditions in a controlled environment.