Item – Thèses Canada

​
Numéro d'OCLC
1013747753
Lien(s) vers le texte intégral
Exemplaire de BAC
Auteur
Lawlor, Sean,
Titre
Traffic estimation and detection methods utilizing automatic vehicle identification systems
Diplôme
Ph. D. -- McGill University, 2017
Éditeur
[Montreal] : McGill University Libraries, [2017]
Description
1 online resource
Notes
Thesis supervisor: Michael Rabbat (Supervisor).
Includes bibliographical references.
Résumé
"Traffic estimation and detection methods have been used for decades to study the roads in urban environments. These road networks and the traffic patterns present on them have traditionally been studied with point-sensor systems such as inductive loops, which record the number of vehicles on a road segment. However in recent time, advances in automatic vehicle identification (AVI) sensors have allowed for a more advanced sensor deployement on these urban roads. These sensors record a unique identifier for a vehicle at each sensor location allowing the vehicles to be \textit{tracked} over time. This thesis presents three topics utilizing data from AVI data to perform a series of tasks ranging from convoy detection to estimation of the traffic flow on an urban road network to estimation of the origin-destination (OD) patterns of travellers on a road network. In the first article, we present a method for identifying vehicles which appear to be traveling in dependent patterns through a sensor network deployed in an urban road environment. The next article looks at expanding the model for nominal traffic to allow for time-varying changes in the traffic as the day progresses. Finally we present a method in the last article which recreates an OD matrix from a stream of AVI data into a time-varying mixture model of the OD matrices present in the road network. The presented methods have applications ranging from law enforcement (for convoy detection), to emergency evacuation management (time-varying traffic pattern estimation), to city planning (estimation of time-varying OD matrices). The collection of methods which are presented in this thesis enrich the field of traffic engineering by allowing models which are only dependent on the data instead of prior biasing information as well as having applications in real-time environments. In the three manuscripts presented, we lay out the analytical methods for detection as well as estimation. We then analyze the algorithms' performance on real and simulated data throughout this work."--
Autre lien(s)
digitool.Library.McGill.CA
escholarship.mcgill.ca
escholarship.mcgill.ca
Sujet
Electrical and Computer Engineering