Date of Award

Spring 2020

Document Type

Thesis

Publisher

Santa Clara : Santa Clara University, 2020.

Departments

Civil, Environmental and Sustainable Engineering; Electrical and Computer Engineering; Computer Science and Engineering

First Advisor

Yi Fang

Second Advisor

Rachel He

Third Advisor

Sarah Kate Wilson

Abstract

A Python-based machine learning algorithm was designed for the function of the traffic signal controller. Traffic signal controllers are the decision-making component within a traffic control box. Signal controllers determine when and which traffic lights transition from one phase to the next. The deep Q-learning algorithm designed in this project looked to decrease average vehicle delay, the expected amount of stoppage the average vehicle should expect, by a minimum of 10%. This reduction in vehicle wait times will have a noticeable impact on vehicle emissions created by interrupted vehicle flow, making the signalized intersection system more environmentally friendly. On top of these performance-based metrics, the design provided by this project can be implemented theoretically in easily available and cost-reasonable hardware, allowing transportation authorities to save considerable amounts of taxpayer money on the designed product.

Download
COinS