Doctoral Dissertations
Date of Award
12-2024
Degree Type
Dissertation
Degree Name
Doctor of Philosophy
Major
Civil Engineering
Major Professor
Asad J. Khattak
Committee Members
Russell Zaretzki, Subhadeep Chakraborty, Candace Brakewood
Abstract
Automated Vehicles (AVs) promise to improve road safety by reducing human errors. However, AVs face safety-related issues that need to be addressed, including ensuring safe interactions with vulnerable road users, e.g., pedestrians, especially at night, managing disengagements (vehicle control shifts from automatic to manual), addressing AV crashes, and managing complex scenarios beyond the AV capabilities, i.e., edge cases. This dissertation addresses these gaps by analyzing the safety concerns of AV operation, utilizing real-world data such as controlled experiments from the Insurance Institute for Highway Safety, California Department of Motor Vehicles disengagement data, and National Highway Traffic Safety Administration AV crash data. This dissertation investigates the effectiveness of Pedestrian Automatic Emergency Braking (P-AEB) in low-level AVs at night and disengagements, intersection crashes (where most crashes occur), and edge-case crashes in high-level AVs. The findings reveal that P-AEBs reduce vehicle-pedestrian crash risks, avoiding crashes in 64% of the tests. However, crashes highlight the need for advanced headlight technology, sensor fusion, improved lateral motion detection algorithms, and enhancements for larger, heavier vehicles and Electric Vehicles (EVs). Humans (safety drivers) initiate most disengagements (88.02%). AV-initiated disengagements are more likely in EVs, older vehicles, and SUVs/vans and less likely in AVs with higher annual vehicle miles traveled. In analyzing AV crashes at intersections, where non-injury crashes predominate, applying the Synthetic Minority Over-sampling Technique improves the injury classification models' predictive accuracy. Further, the absence of safety drivers, precrash disengagements, and Crash Partners' (CPs') unlawful behavior are associated with higher injury probabilities at intersections. Investigating edge cases reveals that human actions contribute to 60% of these cases, and injury rates are higher in edge-case crashes. The main scenarios for edge cases include CPs' unexpected behaviors, absence of safety drivers, precrash disengagements, and unusual events, e.g., unexpected obstacles, unclear road markings, and sudden changes in traffic flow. Overall, the findings indicate that AVs require enhanced interactions with non-compliant road users, infrastructural anomalies, vehicle design, sensor functionality, and software algorithms. This research offers insights for vehicle manufacturers and policymakers, guiding future AV design, testing, and deployment standards to ensure safer integration of AVs into road transportation systems.
Recommended Citation
Moradloo, Nastaran, "THE ROLE OF AUTOMATED VEHICLES IN ENHANCING ROAD SAFETY: A COMPREHENSIVE EVALUATION OF OPERATIONAL SAFETY CHALLENGES IN MIXED-TRAFFIC ENVIRONMENT. " PhD diss., University of Tennessee, 2024.
https://trace.tennessee.edu/utk_graddiss/11375