Skip to Main Content
Frequently Asked Questions
Submit an ETD
Global Search Box
Need Help?
Keyword Search
Participating Institutions
Advanced Search
School Logo
Files
File List
thesis_stoddart_FinalCopy.pdf (3.12 MB)
ETD Abstract Container
Abstract Header
Computer Vision Techniques for Automotive Perception Systems
Author Info
Stoddart, Evan
Permalink:
http://rave.ohiolink.edu/etdc/view?acc_num=osu1555357244145006
Abstract Details
Year and Degree
2019, Master of Science, Ohio State University, Electrical and Computer Engineering.
Abstract
Computer vision is widely used in modern robotics and automated machinery due to its highly diverse ability to recognize patterns, track objects, and map environments. Over the last decade, computer vision has also become an integral part in advanced driver assistance systems (ADAS) which assist drivers in operating motor vehicles. Modern ADAS use sensors, such as cameras, to characterize the environment and either alert the driver or take control to lessen the severity of, and even prevent, accidents. This work discusses a vision-based, SAE Level 0 ADAS system designed for a collegiate automotive design competition. The experimental design and setup will introduce the custom, hybrid-electric ego-vehicle, used as a prototype deployment vehicle for the system, as well as the components and software practices. Next, the computer vision techniques used to detect and track target vehicles, will be discussed. This is structured chronologically in the stages followed to make a camera-based system. This included component modeling and calibration, image processing and computer vision operations, and application-focused feature extraction techniques used in lane line identification, vehicle classification and localization, and vehicle tracking. All software developed for this system used the OpenCV library for Python and C++. This library is easy to use and well-documented, making it a convenient resource for a plethora of vision-based algorithms. Validation techniques will also be discussed in terms of software-in-the-loop, hardware-in-the-loop, and in-vehicle testing. The system placed 4th of 16 schools in the ADAS Evaluation Event at the end of year competition. The performance of the system is discussed. The performance results were sufficient for the needs of the competition; however, additional reliability would be desired for commercial use. The concepts discussed in this work, however, show promise towards implementing a similar system by following the systems-level development and validation practices discussed. This work serves as a template for the design and implementation of a vision-based perception system with relevance to not only driver assistance, but autonomous vehicles and robotics as well.
Committee
Shawn Midlam-Mohler (Advisor)
Giorgio Rizzoni (Committee Member)
Pages
97 p.
Subject Headings
Automotive Engineering
;
Computer Engineering
;
Computer Science
;
Engineering
Keywords
ADAS
;
computer vision
;
machine learning
Recommended Citations
Refworks
EndNote
RIS
Mendeley
Citations
Stoddart, E. (2019).
Computer Vision Techniques for Automotive Perception Systems
[Master's thesis, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1555357244145006
APA Style (7th edition)
Stoddart, Evan.
Computer Vision Techniques for Automotive Perception Systems.
2019. Ohio State University, Master's thesis.
OhioLINK Electronic Theses and Dissertations Center
, http://rave.ohiolink.edu/etdc/view?acc_num=osu1555357244145006.
MLA Style (8th edition)
Stoddart, Evan. "Computer Vision Techniques for Automotive Perception Systems." Master's thesis, Ohio State University, 2019. http://rave.ohiolink.edu/etdc/view?acc_num=osu1555357244145006
Chicago Manual of Style (17th edition)
Abstract Footer
Document number:
osu1555357244145006
Download Count:
1,775
Copyright Info
© 2019, all rights reserved.
This open access ETD is published by The Ohio State University and OhioLINK.