Date of Award
8-19-2024
Document Type
Open Access Thesis
Department
Mechanical Engineering
First Advisor
Nikolaos Vitzilaios
Abstract
Given the pivotal role of the railroad industry in modern transportation and the potential risks associated with track malfunctions, the inspection and maintenance of railroad tracks emerges as a critical concern. While existing solutions excel in performing accurate measurements and detection, they often rely on large, expensive, and time-consuming platforms for inspections. This project, however, seeks to solve the same problem with the use of an unmanned aerial vehicle (UAV), significantly reducing time and cost while maintaining detection capabilities. In particular, this solution is ideal for large-scale, high-level inspections following major events such as floods [6], hurricanes [7] or earthquakes [29]. In such cases, UAVs offer a more efficient solution. Moreover, UAVs can still fulfill many additional inspection needs achieved by current platforms. Hence, this project focuses on developing, implementing, and testing a fully functional, vision-based, autonomous track-following system for UAVs. The creation of a cutting-edge track detection algorithm, TrackNet, is used to identify and interpret railroad tracks from the video stream of an onboard camera. This system is then seamlessly integrated with a customized DJI Matrice 100 UAV to detect and follow railroads in real-time. Notably, this system operates independently of external sensors such as GPS, thanks to its utilization of advanced computer vision techniques. Two distinct approaches utilizing differing camera configurations were developed, tested, and compared. Both systems were found to successfully detect and follow railroad tracks 300 meters in length containing curved and straight sections. The first approach required a forward-facing camera and detected the vanishing point of the track as a control reference. The second approach required a downward-facing camera and detected the center line of the track to be used as a control reference. These two systems were developed and improved to achieve a average track position errors of 1.9766 meters and 2.0342 meters for the forward-facing approach and the downward-facing approach respectively. Utilizing this system, a UAV can autonomously detect and follow railroad tracks, establishing the fundamental framework upon which various inspection algorithms can be developed to suit their specific applications.
Rights
© 2024, Keith Michael Lewandowski
Recommended Citation
Lewandowski, K. M.(2024). UAV-Based Tracking and Following of Railroad Lines. (Master's thesis). Retrieved from https://scholarcommons.sc.edu/etd/7799