Driver Car-following Behavior Simulation using Fuzzy Rule-based Neural Network

This paper proposes a rule-based car-following model to simulate the driver decision process and model the associated vehicle longitudinal action in the car-following regime. In order to analyze individual driver characteristics and extract driving behavior rules, a fuzzy rule based neural network is constructed with the objective of presenting driver action rules under the associated traffic states. The driving rules are calibrated using vehicle trajectory data from the car-following situations of one driver. An artificial technique, reinforcement learning (RL), is applied in the fuzzy rule calibration. Vehicle longitudinal actions are estimated and used as the output of this model. The simulated vehicle actions are compared to the naturalistic data. In the proposed methodology, the naturalistic traffic state and driving actions are extracted from the Naturalistic Truck Driving Study (NTDS) database provided by the Virginia Tech Transportation Institute (VTTI). Driving actions were recorded in instrumented vehicles that have been equipped with specialized sensor, processing, and recording equipment. Car-following situations are extracted by pre-defined criteria.