Intelligent cyclist crash modelling to predict the riskiest infrastructure type

Research output: Contribution to conferencePaperpeer-review

Abstract

The transportation system's present need is to increase the cycling mode share, which can only be achieved by improving its safety. This paper aims to develop a predictive intelligent safety modelling framework for the riskiest cyclist infrastructure based upon the prevalent environment, traffic flow conditions, and specific users. The northeast of England is selected as a case study. A hybrid methodological framework using statistical and machine learning is proposed: a) Crash data collection, b) Predictive model and c) Governing variable analysis model. Firstly, the details of each crash in the study area from 2005-2014 are evaluated. After data cleaning, a base input crash file is developed, used as an input in the data learning model to develop the predictive models. Finally, governing variable analysis is performed to understand their impact on safety. Two accurate infrastructure predictive models are developed with a mean accuracy of 94%. Based on the results obtained, we show that the most critical variables affecting an infrastructure's safety are riders age, gender, environmental conditions, sudden change in the road hierarchy, and the traffic flow regime. It is found that the adverse environmental conditions and different traffic flow regimes complicate the cyclist interactions, having varied safety implications for different infrastructure types. The riskiest environmental conditions are compounded by the prevalent traffic flow regime and present a significant safety hazard for the cyclist. The traffic flow regime poses a varying level of risk to the cyclist to which riders belonging to different genders react differently. The study results help develop a better understanding of risk variation for different infrastructure types and predict the riskiest infrastructure type. It is expected that this study's results will contribute towards better planning of the cyclist infrastructure.
Original languageEnglish
Publication statusPublished - 7 Jul 2021
Event53rd Annual Universities Transport Studies Group Conference - Loughborough University , Loughborough, United Kingdom
Duration: 5 Jul 20216 Jul 2021
http://utsg.net/annual-conference/2021-utsg-conference-papers-extended-abstracts

Conference

Conference53rd Annual Universities Transport Studies Group Conference
Country/TerritoryUnited Kingdom
CityLoughborough
Period5/07/216/07/21
Internet address

Keywords

  • Cycling Safety
  • Intelligent modelling
  • Deep Learning

Fingerprint

Dive into the research topics of 'Intelligent cyclist crash modelling to predict the riskiest infrastructure type'. Together they form a unique fingerprint.

Cite this