Assessing Infrastructure Elements using Automated Object Detection Technique in Smart City Applications
Date of Award
Doctor of Philosophy
Civil and Construction Engineering
Dr. Jun-Seok Oh
Dr. Osama Abudayyeh
Dr. Alvis Fong
object detection automation, trajectory, human behavior, bicycle interactions
Nowadays, road features are becoming more complex leading to more complicated complaints regarding urban environments. Point Cloud Data (PCD) processing is an essential element for detecting objects and analysing human driving behavior to identify the variables defining challenging objects and maneuvers in smart cities. PCDs include a range of processing, including indirect processing (e.g., data converting, cleaning process) and direct process (e.g., pass through elevation filter, statistical outlier removal, normal estimation as well as classification). Static and dynamic object detection and analysis are typically considered the most sophisticated options subsumed under PCDs. They involve direct evaluation of both static and dynamic objects of maneuvers variables for people who use the road with and without a vehicle in line with American with Disability Act (ADA) and trajectory variables of passing distance law for challenging scenarios.
The results of point cloud data evaluations allow government agencies to provide communities with the information necessary to strategically plan transportation infrastructure improvements for people using roads and sidewalks. This two-part study identifies the essential components of sidewalk evaluation, and driver behaviors and reports the degree to which object and driver analysis are aligned with the expert recommended components of the ADA and passing distance law.
The first study explores the feasibility of using both terrestrial laser scanner and open source processing algorithms to develop an approach to automate the evaluation of the alignment of transportation infrastructure with public rights of way. In the second study, a new approach of LIDAR data processing is developed to determine the speed and distance of vehicles approaching and entering the passing zone for bicycles. The model develops a technique for analysing the motorist passing behavior in the natural driving environment using the data collected from real vehicle and bicycle maneuvers.
Mastali, Majid, "Assessing Infrastructure Elements using Automated Object Detection Technique in Smart City Applications" (2019). Dissertations. 3473.