# Development and Evaluation of Two Learning-Based Personalized Driver Models for Car-Following Behaviors

## TL;DR

This paper presents two novel learning-based personalized driver models for car-following behaviors, utilizing naturalistic driving data to improve flexibility and accuracy over traditional physical-based models.

## Contribution

The paper introduces two new learning-based models combining GMM with HMM and GMM with PDF, demonstrating their effectiveness in modeling driver behaviors.

## Key findings

- Both models fit driver data well.
- GMM-PDF model shows higher accuracy with more data.
- Models outperform traditional physical-based approaches.

## Abstract

Personalized driver models play a key role in the development of advanced driver assistance systems and automated driving systems. Traditionally, physical-based driver models with fixed structures usually lack the flexibility to describe the uncertainties and high non-linearity of driver behaviors. In this paper, two kinds of learning-based car-following personalized driver models were developed using naturalistic driving data collected from the University of Michigan Safety Pilot Model Deployment program. One model is developed by combining the Gaussian Mixture Model (GMM) and the Hidden Markov Model (HMM), and the other one is developed by combining the Gaussian Mixture Model (GMM) and Probability Density Functions (PDF). Fitting results between the two approaches were analyzed with different model inputs and numbers of GMM components. Statistical analyses show that both models provide good performance of fitting while the GMM--PDF approach shows a higher potential to increase the model accuracy given a higher dimension of training data.