Skip to Main content Skip to Navigation
Theses

Méthodologie d'exploitation de données non temporelles et temporelles issues d'une étude sur le système conducteur-vehicule-environnement (SCVE) : application à la situation de suivi

Abstract : For 15 years, more and more advanced driving assistance systems have equipped cars. These systems may have positive effects on road security but adverse effects may also appear. That is why, it is important to study natural driver’s behavior to develop efficient and sure systems. In driving studies, non-temporal data (e.g. age, sex, psychological profile…) and temporal data (e.g. speed, longitudinal acceleration…) are recorded. This thesis proposes a methodology to explore a multifactor and multivariate database by using a multidimensional descriptive analysis (Multiple Correspondence Analysis) followed by hypothesis tests to confirm observations. A five stages procedure is followed: Characterization, Scale transformation, Table building, Data analysis application and Presentation of results. To characterize non-temporal data we used fuzzy space windows and for temporal data, we used a fuzzy space-time windowing. To study driver’s behavior in longitudinal regulation, especially in car following situations, a study of 250 km on real road with 123 participants driving an instrumented car was realized. More than 12000 car following situations were recorded. Cognitive and personality profiles were described with neuropsychological and personality tests. We first analyzed non-temporal data to describe the population on a psychological way. Then, we focused on temporal data to describe car following situation and effects of environmental factors (meteorological conditions, traffic…) on driver’s behavior. Finally, in a last analysis, psychological profile of a participant and his driving behavior were related in order to determine the potential effects of personality and cognition on road behavior. Observations made in MCA plane were illustrated with graphical models or confirmed with hypothesis tests.
Document type :
Theses
Complete list of metadatas

https://hal-uphf.archives-ouvertes.fr/tel-03046076
Contributor : Julie Cagniard <>
Submitted on : Tuesday, December 8, 2020 - 11:48:26 AM
Last modification on : Wednesday, December 9, 2020 - 3:28:39 AM

File

2009VALE0024_DUBART_DELPHINE.p...
Files produced by the author(s)

Identifiers

  • HAL Id : tel-03046076, version 1

Collections

Citation

Delphine Dubart. Méthodologie d'exploitation de données non temporelles et temporelles issues d'une étude sur le système conducteur-vehicule-environnement (SCVE) : application à la situation de suivi. Automatique. Université de Valenciennes et du Hainaut-Cambrésis, 2009. Français. ⟨NNT : 2009VALE0024⟩. ⟨tel-03046076⟩

Share

Metrics

Record views

22

Files downloads

6