A Microscopic Lane-Changing Model For Driver-Vehicle Agent

Lane-changing model is one of the most fundamental models in microscopic traffic simulation, which describes adjustment and implementation of driving goals through interaction with the traffic environment such as the speed and gap of its surrounding vehicles according to driver's personal characteristics. Lane-changing behavior directly affects the traffic conditions in urban road network, so that some traffic phenomena emerges with drivers'lane-changing behavior. Lane-changing model will have great influences on the reliability of the microscopic traffic simulation system. Route selection is one of the vital problems faced by any driver under dynamic traffic environment. It is used to find a path algorithm for driver's route selection in local dynamic road networks.This dissertation aims to study and implement a microscopic lane-changing model for urban traffic microscopic simulation system, as well as the driver's route selection. The work of this dissertation is as follows:(1) The behavioral characteristics of drivers and vehicles are studied at first and a Driver-Vehicle Agent hybrid architecture is presented based on Agent theory, the communication mechanism of the Driver-Vehicle Agents is also addressed.(2) A fuzzy lane-changing model is proposed for Driver-Vehicle Agent. Fuzzy inference is used to model both the discretionary and mandatory lane-changing behavior. After the careful experimental data collection and the simulation platform is constructed, a simulation experiment is conducted to validate this lane-changing model.(3) Dynamic route selection of the Driver-Vehicle Agent is further studied. By considering the direction information and the actual dynamic traffic environment, a local path algorithm based on A* algorithm is developed to search a new route for Driver-Vehicle Agent's route selection. Lastly a simulation experiment is conducted to validate the algorithm proposed in this dissertation.