A hybrid approach for identifying factors affecting driver reaction time using naturalistic driving data

Nasim Arbabzadeh, Mohsen Jafari, Mohammad Jalayer, Shan Jiang, Mohamed Kharbeche

Research output: Contribution to journalArticle

Abstract

The National Transportation Safety Board (NTSB) estimates that 80% of the deaths and injuries resulting from rear-end collisions could be prevented by the use of advanced collision avoidance systems. While autonomous or higher-level vehicles will be equipped with this technology by default, most of the vehicles on our roadways will lack these advances, so rear-end crashes will dominate accident statistics for many years to come. However, a simple and cost-effective in-vehicle device that uses predictive tools and real-time driver-behavior and roadway data can significantly reduce the likelihood of these crashes. In this paper, we propose a hybrid physics/data-driven approach that can be used in a kinematic-based forward-collision warning system. In particular, we use a hierarchical regularized regression model to estimate driver reaction time based on individual driver characteristics, driving behavior, and surrounding driving conditions. This personalized reaction time is input into the Brill's one-dimensional car-following model to calculate the critical distance for collision warning. We use the Second Strategic Highway Research Program (SHRP-2)'s Naturalistic Driving Study (NDS) data, the largest and most comprehensive study of its kind, to model driver brake-to-stop response time. The results show that the inclusion of driver characteristics increases model precision in predicting driver reaction times.

Original languageEnglish (US)
Pages (from-to)107-124
Number of pages18
JournalTransportation Research Part C: Emerging Technologies
Volume100
DOIs
StatePublished - Mar 1 2019

Fingerprint

driver
Alarm systems
Collision avoidance
Brakes
accident statistics
Accidents
Kinematics
Railroad cars
Physics
traffic behavior
Statistics
physics
time
inclusion
death
Costs
regression
lack
costs

Cite this

@article{083abe617e364d999c934916ae844aa8,
title = "A hybrid approach for identifying factors affecting driver reaction time using naturalistic driving data",
abstract = "The National Transportation Safety Board (NTSB) estimates that 80{\%} of the deaths and injuries resulting from rear-end collisions could be prevented by the use of advanced collision avoidance systems. While autonomous or higher-level vehicles will be equipped with this technology by default, most of the vehicles on our roadways will lack these advances, so rear-end crashes will dominate accident statistics for many years to come. However, a simple and cost-effective in-vehicle device that uses predictive tools and real-time driver-behavior and roadway data can significantly reduce the likelihood of these crashes. In this paper, we propose a hybrid physics/data-driven approach that can be used in a kinematic-based forward-collision warning system. In particular, we use a hierarchical regularized regression model to estimate driver reaction time based on individual driver characteristics, driving behavior, and surrounding driving conditions. This personalized reaction time is input into the Brill's one-dimensional car-following model to calculate the critical distance for collision warning. We use the Second Strategic Highway Research Program (SHRP-2)'s Naturalistic Driving Study (NDS) data, the largest and most comprehensive study of its kind, to model driver brake-to-stop response time. The results show that the inclusion of driver characteristics increases model precision in predicting driver reaction times.",
author = "Nasim Arbabzadeh and Mohsen Jafari and Mohammad Jalayer and Shan Jiang and Mohamed Kharbeche",
year = "2019",
month = "3",
day = "1",
doi = "https://doi.org/10.1016/j.trc.2019.01.016",
language = "English (US)",
volume = "100",
pages = "107--124",
journal = "Transportation Research Part C: Emerging Technologies",
issn = "0968-090X",
publisher = "Elsevier Limited",

}

A hybrid approach for identifying factors affecting driver reaction time using naturalistic driving data. / Arbabzadeh, Nasim; Jafari, Mohsen; Jalayer, Mohammad; Jiang, Shan; Kharbeche, Mohamed.

In: Transportation Research Part C: Emerging Technologies, Vol. 100, 01.03.2019, p. 107-124.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A hybrid approach for identifying factors affecting driver reaction time using naturalistic driving data

AU - Arbabzadeh, Nasim

AU - Jafari, Mohsen

AU - Jalayer, Mohammad

AU - Jiang, Shan

AU - Kharbeche, Mohamed

PY - 2019/3/1

Y1 - 2019/3/1

N2 - The National Transportation Safety Board (NTSB) estimates that 80% of the deaths and injuries resulting from rear-end collisions could be prevented by the use of advanced collision avoidance systems. While autonomous or higher-level vehicles will be equipped with this technology by default, most of the vehicles on our roadways will lack these advances, so rear-end crashes will dominate accident statistics for many years to come. However, a simple and cost-effective in-vehicle device that uses predictive tools and real-time driver-behavior and roadway data can significantly reduce the likelihood of these crashes. In this paper, we propose a hybrid physics/data-driven approach that can be used in a kinematic-based forward-collision warning system. In particular, we use a hierarchical regularized regression model to estimate driver reaction time based on individual driver characteristics, driving behavior, and surrounding driving conditions. This personalized reaction time is input into the Brill's one-dimensional car-following model to calculate the critical distance for collision warning. We use the Second Strategic Highway Research Program (SHRP-2)'s Naturalistic Driving Study (NDS) data, the largest and most comprehensive study of its kind, to model driver brake-to-stop response time. The results show that the inclusion of driver characteristics increases model precision in predicting driver reaction times.

AB - The National Transportation Safety Board (NTSB) estimates that 80% of the deaths and injuries resulting from rear-end collisions could be prevented by the use of advanced collision avoidance systems. While autonomous or higher-level vehicles will be equipped with this technology by default, most of the vehicles on our roadways will lack these advances, so rear-end crashes will dominate accident statistics for many years to come. However, a simple and cost-effective in-vehicle device that uses predictive tools and real-time driver-behavior and roadway data can significantly reduce the likelihood of these crashes. In this paper, we propose a hybrid physics/data-driven approach that can be used in a kinematic-based forward-collision warning system. In particular, we use a hierarchical regularized regression model to estimate driver reaction time based on individual driver characteristics, driving behavior, and surrounding driving conditions. This personalized reaction time is input into the Brill's one-dimensional car-following model to calculate the critical distance for collision warning. We use the Second Strategic Highway Research Program (SHRP-2)'s Naturalistic Driving Study (NDS) data, the largest and most comprehensive study of its kind, to model driver brake-to-stop response time. The results show that the inclusion of driver characteristics increases model precision in predicting driver reaction times.

UR - http://www.scopus.com/inward/record.url?scp=85060227612&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85060227612&partnerID=8YFLogxK

U2 - https://doi.org/10.1016/j.trc.2019.01.016

DO - https://doi.org/10.1016/j.trc.2019.01.016

M3 - Article

VL - 100

SP - 107

EP - 124

JO - Transportation Research Part C: Emerging Technologies

JF - Transportation Research Part C: Emerging Technologies

SN - 0968-090X

ER -