Behavior-based Predictive Safety Analytics

Pilot study

Report (2019)

Authors

F.A. Dreger Human-Robot Interaction - Mechanical, Maritime and Materials Engineering
J. C.F. de Winter Human-Robot Interaction - Mechanical, Maritime and Materials Engineering
J.C.F. de Winter Human-Robot Interaction - Mechanical, Maritime and Materials Engineering
Research Group
Human-Robot Interaction (Mechanical, Maritime and Materials Engineering) (TU Delft)
Copyright: © 2019 Johan Engström, Andrew Miller, Wenyan Huang, Susan Soccolich, Sahar Ghanipoor Machiani, Arash Jahangiri, F.A. Dreger, J.C.F. de Winter
More Info
expand_more

Published Date

2019

Language

English

Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Faculty

Mechanical, Maritime and Materials Engineering

Department

Cognitive Robotics

Research Group

Human-Robot Interaction

Abstract

This report gives an overview of the main findings from the Behavior-based Predictive Safety Analytics – Pilot Study project. The main objective of the project was to investigate the possibilities of developing statistical models predicting individual driver crash involvement based on individual driving style, demographic and behavioral history variables, using large sets of naturalistic driving data. The project was designed as a pilot project with the objective of providing the basis for a future more comprehensive research effort. Based on Second Strategic Highway Research Program (SHRP2) data, a subset of behavior and crash data including 2,458 drivers was created for analysis. The data were analyzed to investigate to what extent these drivers were differentially involved in crashes and near crashes, to what extent this was associated with individual characteristics, and if it is possible to predict individual drivers’ crash and near crash involvement based on variables representing individual characteristics. The results clearly demonstrated the presence of differential crash and near crash involvement and showed significant associations between enduring personal factors and crash involvement. Moreover, logistic regression and random forest classifiers were relatively successful in predicting crash and near crash involvement based on individual characteristics, but the ability to specifically predict involvement in crashes was more limited.

Files

02_020_Final_Research_Report_F... (pdf)
(pdf | 7.13 Mb)
Unknown license