2026-05-20T21:11:54Zhttps://keep.lib.asu.edu/oai/request

oai:keep.lib.asu.edu:node-1982012024-12-23T18:01:48Zoai_pmh:repo_items

198201 https://hdl.handle.net/2286/R.2.N.198201 http://rightsstatements.org/vocab/InC/1.0/ All Rights Reserved 2024 2026-12-01T11:30:18 71 pages Masters Thesis Academic theses Text eng Saravanan, Nithish Kumar Zhao, Junfeng Suo, Dajiang Yang, Yezhou Arizona State University Partial requirement for: M.S., Arizona State University, 2024 Field of study: Engineering Perception is a key component in Autonomous Vehicles (AVs). However, sensors mounted on the AVs often encounter blind spots due to occlusion caused by nearby vehicles, infrastructure, or surrounding elements. While recent advancements in planning and control algorithms help AVs react to sudden object appearances from blind spots at low speeds and less complex scenarios, challenges remain at high speeds and complex intersections. According to a Waymo report, two out of eight intersection accidents are due to occlusion. Vehicle-to-Infrastructure (V2I) technology promises to enhance scene representation for Connected and Automated Vehicles (CAVs) in complex intersections, providing sufficient time and distance to react to adversary vehicles violating traffic rules. Most existing methods for infrastructure based vehicle detection and tracking rely on LiDAR or sensor fusion methods, such as LiDAR-Camera. Although LiDAR provides accurate spatial information, the sparsity of point cloud data limits its detection range. Furthermore, the absence of LiDAR at every intersection increases the cost of implementing V2I technology. To overcome these problems, this thesis aims to utilize existing monocular traffic cameras at road intersections for 3D object detection, combining results with onboard systems using an asynchronous late fusion method to enhance scene representation. Additionally, the proposed framework in this thesis provides a time delay compensation module to compensate for the processing and transmission delay from the Roadside Unit (RSU). Finally, the effectiveness of the proposed V2I framework is evaluated by recreating and validating a scenario similar to the one reported in Waymo's Public Safety Performance Data, where their automated driving system failed to detect an adversary vehicle due to occlusion, leading to a crash. Robotics Asynchronous late object fusion Connected and Automated Vehicle Monocular traffic camera Time delay compensation Monocular Traffic Camera-Based Vehicle to Infrastructure Framework Using Time Delay Compensation and Late Fusion