Analysis of Autonomous Vehicles Accidents Based on DMV Reports

Analysis of the autonomous vehicles accidents is one of the best ways to provide support for improving auto driving systems. This paper uses the publicly available dataset from the California DMV to deeply study the problem of current auto driving systems. The original data is processed according to its features. Some new definition has been made to clearly analyze the precise reason for vehicle accidents under auto driving mode. The result indicates that rear-ended collision is the main type of accident. Based on some further analysis, possible causes of the accident related to the inadequate auto driving system are presented in the paper.