Abstract

A key factor leading to the inefficiency of transportation systems is the lack of real-time, fine-grained traffic analytics. An average motorist in the United States spends a significant amount of time looking for parking spots during their daily commute. Existing systems for traffic analytics are either not scal- able to large metropolitan areas or require a human-in-the-loop. This project focuses on automating the detection of fine-grained traffic analytics. Lever- aging on-board stereo cameras, Autowaze utilizes a crowd-sourced strategy to take time-windowed snapshots of the road containing 3D map points of the environment. These snapshots are used to extract changes in the environment which are uploaded to a central cloud server responsible for inferring traf- fic analytics such as vacant parking spots. Evaluations show that Autowaze correctly predicts the occupancy status of a parking spot 89% of the time. Moreover, the use of this system can lead to great reductions in search time for motorists looking for available parking as well as increased fuel efficiency and cost savings.

Library of Congress Subject Headings

Traffic engineering--Data processing; Parking facilities--Data processing; Crowdsourcing

Publication Date

5-9-2024

Document Type

Thesis

Student Type

Graduate

Degree Name

Computer Science (MS)

Department, Program, or Center

Computer Science, Department of

College

Golisano College of Computing and Information Sciences

Advisor

Fawad Ahmad

Advisor/Committee Member

Zachary Butler

Advisor/Committee Member

M. Mustafa Rafique

Campus

RIT – Main Campus

Plan Codes

COMPSCI-MS

Download

Share

COinS