Keywords

Automated Vehicles, Crash Analysis, Matched case-control study, Random parameters multinomial logit models, Quasi-Induced Exposure, Bayesian Hierarchical Random-parameter Model

Abstract

Automated Vehicles (AV) are increasingly becoming a reality, offering major road safety improvements and advancements through the development of Advanced Driver Assistance Systems (ADAS) and Automated Driving Systems (ADS). Although research on AVs is progressing, several key gaps remain. There is a lack of sufficient matched data between AV and HDV crashes to compare their differences, along with a need for more detailed datasets to analyze ADS and ADAS crash severity and assess ADAS safety effectiveness. First, the research investigates overall trends within the AV dataset by utilizing 1,099 ADS, 1,001 ADAS and 35,113 HDV crash data. including ADS (Society of Automotive Engineers (SAE) Level 4) and ADAS (SAE Level 2). And then, a matched case-control design is employed to compare differences between ADS and Human-Driven Vehicle (HDV). Second, this research focuses on analyzing the injury severity of 1,001 ADAS and 548 ADS crashes. Two random parameters multinomial logit models with heterogeneity in the means are considered to analyze the contributing factors of injury severity for the ADAS and ADS. Lastly, this research assesses the safety effectiveness of ADAS using 3,549 ADAS-equipped and 4.51 million non-ADAS vehicles involved in two-vehicle crashes across 34 models and 41 brands, with these ADAS always activated. The Quasi-Induced Exposure (QIE) is applied to assess the safety effectiveness across various conditions. Also, a series of Bayesian hierarchical models are compared and the best fit Bayesian Hierarchical Random-Parameter Model (BHRPM) considering unobserved heterogeneity is selected to investigate the determinants influencing the ADAS-equipped vehicle responsibility in a crash. iv Through these objectives, this research assesses ADAS and ADS vehicle safety using real-world crash data, offering insights to guide deployment strategies and future advancements in AV safety.

Completion Date

2024

Semester

Fall

Committee Chair

Mohamed Abdel-Aty

Degree

Doctor of Philosophy (Ph.D.)

College

College of Engineering and Computer Science

Department

Dept. of Civil, Environmental & Construction Engineering

Format

PDF

Identifier

DP0029697

Document Type

Thesis

Campus Location

Orlando (Main) Campus

Download

Share

COinS