Back to search

University of Patras

Ανάπτυξη πλαισίου επικινδυνότητας στην κυκλοφοριακή ροή αυτοκινητοδρόμων

Abstract

dc:description

Road transport is a major societal and economic activity. Despite its importance, road transport raises a number of challenges that have to be dealt with. Road safety is the most important of them. Fatal road accidents lead to the death of 1.35 million people annually, while they are the 8th cause of death worldwide and the 1st cause of death for the people who belong to the age group 5-29 year-old.Road safety and traffic flow theory constitute an important research and study field. In the framework of this thesis risk conditions in freeways are studied microscopically. Two traffic incidents are considered as risk conditions: shockwave and forced lane change. These two incidents are non-catastrophic incidents, which are used as crash surrogates. The research studies about these incidents have focused on the characteristics of the phenomena, their interpretation, their mechanisms and laws as well as their impact on traffic flow. There is though a lack of research on the causes generating these incidents and on their description. To better understand the risk conditions a large dataset of vehicle trajectories has been developed. The extracted trajectories correspond to vehicles involved in two traffic incidents, shockwave and forced lane change respectively. These incidents are spotted after watching several hours of traffic video recordings. Moreover, trajectories of vehicles not involved in incidents are developed. The trajectories dataset, which was developed manually, is different from the widely used dataset of NGSIM, as it focuses on incidents instead of continuous traffic and it offers the possibility to watch the video recordings of the traffic which correspond to the trajectories. After the calibration of the trajectories, the traffic variables which describe the traffic flow microscopically are selected. For the incidents the value of the variables are calculated for the start time of the incidents. For forced lane changes, start time is considered the moment when the vehicles starts maneuvering to change lane. For shockwaves, start time is considered the moment when the driver starts braking. For incident-free flow, the variables are calculated at random moments. For the most efficient study of the data, 4 populations are generated: population A, which includes all incidents -both shockwaves and forced lane changes- and incident-free flow, population B which includes shockwaves and incident-free flow, population C which includes forced lane changes and incident-free flow, and population D which includes only incidents, shockwaves and forced lane changes. The values of the variables, which have been developed based on the dataset, are clustered by employing two clustering algorithms based on cost function minimization: K-means algorithm and Fuzzy c-means algorithm. To implement the algorithms: a) the number of the clusters is determined; in all four populations it is well-known that there are two clusters, b) the couple of variables used is defined, c) the algorithms are implemented, and d) the algorithm results are evaluated by using two indexes: the detection rate and the false alarm rate. Clustering demonstrated the significance of space headway to make the distinction between risk conditions or absence of risk.Discrete choice models (Logit) are developed to study the risk conditions from the perspective of the driver, based on the driver’s choices. More specifically, in the case of existing or non-existing risk, two discrete choices are defined for the driver: 1. The driver does not change his/her driving behavior 2. The driver implements mitigation actionFor population Α, the driver’s discrete choices are modified as follows: 1) the driver does not change his/her driving behavior and 2) the driver changes his/her driving behavior. For population B, the driver’s discrete choices are modified as follows: 1) the driver does not change his/her driving behavior and 2) the driver decelerates. For population C, the two discrete choices are modified as follows: 1) the driver does not change his/her driving behavior and 2) the driver changes lane. For population D, which includes only incidents, the discrete choices are modified as follows: 1) the driver decelerates 2) the driver changes lane. A control system for risk management in freeways at microscopic level is developed. The system utilizes prior knowledge of risk conditions in freeways in order to limit risk. The suggested control system follows the movement in a lane of freeway in which all vehicles are equipped with sensors and are able to transmit information about their position y and their velocity v at any time, and are also able to receive the same information from their preceding vehicle. The state variables of the system are the change of the position (Δx) covered by each one of two consecutive vehicles N, N+1 at a time interval equal to the sampling time. The control scheme suggests that we ought to measure two consecutive vehicles space headway in real time, and classify it to a risk level. If a vehicle finds itself at risk conditions, then the control system gets activated and the following vehicle’s velocity gets reduced so that the space headway gets increased. Suggestions for future research refer to the need to extend the study of risk conditions to other traffic phenomena such as the pseudo lane-change, as well as to other emerging modes of transport such as micromobility.

Degree

thesis:*
Grantor dc:publisher
University of Patras
Year dc:date
2019

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Gioldasis, Christos
  • Γιολδάσης, Χρήστος

Subjects

dc:subject × 12

Rights

Language dc:language
gre

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/47146

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Gioldasis, Christos; Γιολδάσης, Χρήστος. Ανάπτυξη πλαισίου επικινδυνότητας στην κυκλοφοριακή ροή αυτοκινητοδρόμων. University of Patras, 2019. http://hdl.handle.net/10442/hedi/47146