Modelling Intra-driver Behavioral Adaptation through Risk Sensitivity and Regime Transitions: A Task-difficulty Car-following Model

Abstract

Over the past decade, there has been a growing trend toward integrating human factors (HF) into traffic flow models to better understand the complexities of human behavior and its impact on traffic dynamics. This research seeks to advance this trend by bridging the gap between traditional car-following models and the inherent variability of human driving behavior. By incorporating these elements, our models provide valuable insights into how driver behavior adaptation and risk-taking influence traffic flow dynamics. Specifically, the study proposes a model called Intelligent Driver Model Task Saturation that integrates human behavioral adaptations and risk-taking strategies into a modified version of the established Intelligent Driver Model, enriched by a cognitive layer based on Fuller's Task Capability Interface model. This amalgamation offers a perspective on the interplay between driver behavior adaptation when the driving task saturates and the risk-taking strategy of drivers. When total task demand exceeds task capacity, drivers may adapt their behaviors in accordance with Fuller's risk allostasis theory, aiming to mitigate risk to acceptable levels. The model was calibrated utilizing data from driving simulator scenarios involving both normal and distracted driving and naturalistic data, ensuring its sensitivity to diverse driving behaviors and adaptive behavioral changes. We investigated the model in terms of behavioral soundness and model fitting. Our results demonstrate that the model effectively incorporates endogenous mechanisms to explain both inter- and intra-driver heterogeneities in driving behavior. Additionally, it generates two plausible HF: risk-taking and behavior adaptation. The findings of this study suggest a step forward in achieving a more realistic representation of driving behavior adaptation and risk-taking strategies.