Date of Award

Summer 8-2016

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Computational Modeling & Simulation Engineering

Program/Concentration

Modeling and Simulation

Committee Director

Mecit Cetin

Committee Member

Roland Mielke

Committee Member

Jiang Li

Committee Member

Tamer Nadeem

Abstract

Building sustainable traffic control solutions for urban streets (e.g., eco-friendly signal control) and highways requires effective and reliable sensing capabilities for monitoring traffic flow conditions so that both the temporal and spatial extents of congestion are observed. This would enable optimal control strategies to be implemented for maximizing efficiency and for minimizing the environmental impacts of traffic. Various types of traffic detection systems, such as inductive loops, radar, and cameras have been used for these purposes. However, these systems are limited, both in scope and in time. Using GPS as an alternative method is not always viable because of problems such as urban canyons, battery depletion, and precision errors.

In this research, a novel approach has been taken, in which smartphone low energy sensors (such as the accelerometer) are exploited. The ubiquitous use of smartphones in everyday life, coupled with the fact that they can collect, store, compute, and transmit data, makes them a feasible and inexpensive alternative to the mainstream methods. Machine learning techniques have been used to develop models that are able to classify vehicle movement and to detect the stop and start points during a trip. Classifiers such as logistic regression, discriminant analysis, classification trees, support vector machines, neural networks, and Hidden Markov models have been tested. Hidden Markov models substantially outperformed all the other methods. The feature quality plays a key role in the success of a model. It was found that, the features which exploited the variance of the data were the most effective.

In order to assist in quantifying the performance of the machine learning models, a performance metric called Change Point Detection Performance Metric (CPDPM) was developed. CPDPM proved to be very useful in model evaluation in which the goal was to find the change points in time series data with high accuracy and precision.

The integration of accelerometer data, even in the motion direction, yielded an estimated speed with a steady slope, because of factors such as phone sensor bias, vibration, gravity, and other white noise. A calibration method was developed that makes use of the predicted stop and start points and the slope of integrated accelerometer data, which achieves great accuracy in estimating speed.

The developed models can serve as the basis for many applications. One such field is fuel consumption and CO2 emission estimation, in which speed is the main input. Transportation mode detection can be improved by integrating speed information. By integrating Vehicle (Phone) to Infrastructure systems (V2I), the model outputs, such as the stop and start instances, average speed along a corridor, and queue length at an intersection, can provide useful information for traffic engineers, planners, and decision makers.

Rights

In Copyright. URI: http://rightsstatements.org/vocab/InC/1.0/ This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s).

DOI

10.25777/8fqh-vc04

ISBN

9781369358308

Download

Share

COinS