Personalized Driving Behaviors and Fuel Economy over Realistic Commute Traffic: Modeling, Correlation, and Prediction

TL;DR

This paper develops a personalized driver behavior and fuel consumption model using real traffic data, revealing significant correlations and enabling fuel economy predictions to promote sustainable driving practices.

Contribution

It introduces an integrated microscopic model linking driving behaviors to fuel economy and employs Gaussian Process Regression for personalized fuel consumption prediction.

Findings

Driving behavior differences can cause up to 29% variation in fuel consumption.

Significant correlations exist between individual driving preferences and fuel economy.

The model accurately predicts fuel consumption across diverse traffic and vehicle conditions.

Abstract

Drivers have distinctively diverse behaviors when operating vehicles in natural traffic flow, such as preferred pedal position, car-following distance, preview time headway, etc. These highly personalized behavioral variations are known to impact vehicle fuel economy qualitatively. Nevertheless, the quantitative relationship between driving behaviors and vehicle fuel consumption remains obscure. Addressing this critical missing link will contribute to the improvement of transportation sustainability, as well as understanding drivers' behavioral diversity. This study proposed an integrated microscopic driver behavior and fuel consumption model to assess and predict vehicle fuel economy with naturalistic highway and local commuting traffic data. Through extensive Monte Carlo simulations, significant correlation results are revealed between specific individual driving preferences and fuel economy over drivers' frequent commuting routes. Correlation results indicate that the differences in fuel consumption incurred by various driving behaviors, even in the same traffic conditions, can be as much as 29% for a light-duty truck and 15% for a passenger car. A Gaussian Process Regression model is further trained, validated, and tested under different traffic and vehicle conditions to predict fuel consumption based on drivers' personalized behaviors. Such a quantitative and personalized model can be used to identify and recommend fuel-friendly driving behaviors and routes, demonstrating a strong incentive for relevant stakeholders.