Modeling Lane Changing Behavior using Fuzzy Logic

Abstract

Microscopic traffic simulation is a safe and efficient tool in transport system analysis and management. It can offer a safe, controllable and repeatable environment to study and evaluate the performance of transport network systems under various alternative management options. The model of driver's behavior is a decisive factor for traffic simulation. This paper attempts to present a lane changing model which can represent reality as-near-to-life as possible. Generally, a driver makes decisions based on inexact and linguistic measures of environment state. Hence, it seems worthwhile to establish a direct relationship between the loose linguistic expression of a decision and environment state. Unlike traditional logical systems, fuzzy logic is concerned with imprecise rather than exact modes of reasoning, which plays an important role in the human ability to make decisions in an environment of uncertainty and imprecision. It offers a rigorous and practical technique for manipulating such information originally expressed in a linguistic form. Therefore, fuzzy logic is used to simulate driver's lane changing behavior in this paper. The corresponding fuzzy logic controller is developed in this paper. This fuzzy logic controller is a MISO (Multi-input/Single-Out) system. The input variables include the states of SV, lead vehicle, and vehicles in the target lane and the relative states between these vehicles and environment states. A set of linguistic rules are introduced to model driver's decision-making process. The fuzzy logic controller has three layers for it has multi inputs. The simulation program about this method is written based on VC++ in this paper. The effectiveness of modeling lane changing behavior using fuzzy logic is measured in this paper.