An Investigation into the Safety and Collision Data of Autonomous Vehicles

Abstract

Autonomous vehicles are the future of transportation. Increased safety, accessibility and convenience are just some of the capabilities they can bring to society. The goal of this thesis is to asses safety by analyzing crash data of both autonomous and conventional vehicles. It is imperative to understand how these vehicles interact on the road, as well as, identify any factors correlating to the collisions of the autonomous vehicles. After completing significance tests comparing locations, timing and type of the accidents involving a vehicle in autonomous mode against all of the crash data for the entire state of California, discrepancies were found between the two data sources. One key difference was the proportion of rear-end accidents. Further investigation was required to determine how and why the accident occurred, in addition to, associating fault to each autonomous vehicle crash. The results deemed that autonomous vehicles were not responsible for 97% of the accidents, however, excessive caution from the autonomous vehicles is also a factor. Other drivers on the road are having difficulty anticipating some safety maneuvers of the autonomous vehicles. Autonomous transportation may be having some issues integrating with conventional drivers, but they show great signs of increased safety and will continue to progress moving forward.