Prediction of the Meta-stability Phase through Analysis of Driving Behavior

摘要

Traffic jams are formed in three phases: the free travel phase, the meta-stability phase, in which allows unchanged travel speed with only vehicle density increased, and the traffic jam phase. Therefore, it can be considered that if the meta-stability phase can be detected, forecasting traffic jams becomes possible. Moreover, it can also be considered that drivers unconsciously change their driving behavior based on changes in the surrounding environment. This article proposes a driver model that forecasts traffic jams based on changes in driving behavior and that does not rely on traffic flow monitoring infrastructure. As a result of evaluation in driving simulators, it was understood that the distribution of steering and throttle input frequency changes based on changes in the travel phase. It is possible to distinguish these changes using neural networks, and it is possible to make this into a driver model that forecasts traffic jams. This article will discuss experiments regarding changes in driving behavior in each travel phase, and a driver model that forecasts traffic jams constructed based on analysis of the results of the experiments.