An Exploratory Analysis of Temporal and Spatial Patterns of Autonomous Vehicle Collisions

Abstract

Recent advancements in autonomous vehicle (AV) technology have the potential to reduce road accidents caused by human error. However, to enhance their safety and performance, it is crucial to understand the patterns of AV collisions. This study examines AV collisions by analyzing their temporal and spatial patterns. Based on reports from the California DMV between 2014 and 2022, the analysis reveals that rear-end collisions are the most common type, while incidents involving pedestrians and overturned vehicles are rare. The majority of collisions involve mid-sized vehicles, and AVs are responsible for a minority of accidents. The study also identifies clusters of incidents in San Francisco, San Jose, Los Angeles, and San Diego, with San Francisco having largest concentration. Specific areas within San Francisco, like Mission District, Japantown, Union Square, and North Beach neighborhoods, show high incident rates. These findings highlight safety concerns, and aid in integration of AVs into transportation infrastructure.