ファジィニューラルネットワークモデルによる追従自動車運転理論の実現

概要

論文の詳細を見る
We have already proposed a new road traffic simulation system with a microscopic model for analyzing traffic jams over wide areas (MITRAM). The MITRAM has the microscopic model which consists of Fuzzy Model Vehicle (FMV) with multistage binomial fuzzy logic inference. The FMV describes an individual behavior of vehicle by fuzzy inferences of driving logic. Many membership functions of the FMV are determined with neural network automatically. The existing real FMV's which are developed to simulate all actual driving behaviors have not been acquired in the case that the operation of the actual vehicles was complicated. In this paper, we summarized our previous researches about the modeling of the FMV, and tried to simulate the actual drivings of succeeding vehicles by using two types of the FMV models which have difference in the input parameters. At first, we described the characteristics and the concrete model of the FMV. Then we evaluated time series data of the succeeding vehicles in distance, velocity, and acceleration to compare the simulation results of the two types of the FMV models with the actual driving. From the results we found the FMV models could simulate the actual drivings of succeeding vehicles by means of suitable selection of the input parameters.
日本シミュレーション学会の論文
1999-12-15