A Study on Prediction of Optimal Location for Traffic Enforcement Using Q-Learning

Abstract

The occurrence of traffic accidents in Japan has continued to decrease, but the road traffic environment is expected to change rapidly in the future with the development of next-generation transportation. In this context, there have been many moves in recent years in various fields to use AI (artificial intelligence) to analyze big data and predict events that have been difficult to predict using conventional methods. In the field of traffic safety, the Metropolitan Police Department is also using AI to predict the occurrence of traffic accidents in order to implement effective traffic enforcement activities. In this study, we developed a model that predicts the optimal location for traffic enforcement activities by using Q-Learning, a type of AI. Furthermore, by quantitatively evaluating the deterrent effect of traffic accidents caused by the traffic enforcement activities based on the prediction results, we can obtained knowledge on how to effectively implement enforcement activities in the future.