Banca de DEFESA: Júlia Alves Porto

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
STUDENT : Júlia Alves Porto
DATE: 30/06/2023
TIME: 14:00
LOCAL: Apresentação Híbrida (Sala virtual e sala A1 12/6 - Anexo SG 12)
TITLE:

USE OF CONFLICT DATA FOR ROAD SAFETY ASSESSMENT

KEY WORDS:

Road safety, conflicts, machine learning

PAGES: 69
BIG AREA: Engenharias
AREA: Engenharia de Transportes
SUMMARY:

Traditionally, road safety studies are based on accident data. However, the use of these data causes a series of problems, especially: the time required to obtain the amount of data sufficient to perform the necessary statistical analyses; the lack of reliability; and the difficulty of accessing the data. For these reasons, alternative methods of safety assessment have been used, among which the use of conflict data stands out. A traffic conflict is a traffic interaction that could have resulted in an accident but was not due to evasive action taken by one of the parties involved. The severity of the conflict can be measured by behavioral indicators or by the spatial or temporal proximity between those involved. Observing the occurrence of conflicts, however, is not a simple task and requires a lot of time, in addition to being subject to the subjectivity of the observer. To overcome these difficulties, computer vision codes have been published that allow vehicle tracking and the automatic measurement of conflict indices which allow the conflict severity classification. In this work, a bibliographic review of alternative methods and techniques of applied case studies using conflict data in recent years was made. A study case of conflicts in the Brazilian Federal District was also made using both manual techniques for assessing conflict data, and the software Traffic Intelligence, developed in partnership by researchers of the University of British Columbia and Polytechnique Montréal. This dissertation contributes to the understanding of conflict data assessment, pointing out difficulties still found regarding the lack of standardization on the data collection methods, high subjectivity for manual indicators assessment and considerate error indices using computer vision techniques. These downsides can be overthrown by technological development, specially in the computational engineering field.

BANKING MEMBERS:
Interna - 1963483 - FABIANA SERRA DE ARRUDA
Externo à Instituição - FLÁVIO JOSÉ CRAVEIRO CUNTO - UFC
Presidente - 2492216 - MICHELLE ANDRADE
Interno - 1552603 - PASTOR WILLY GONZALES TACO