The development of automated vehicle (AV) technology is suggesting a promising future of safer and more efficient transportation. However, there are still many challenges in ensuring the operational safety of AVs before their deployment. Scenario-based testing of AVs is an essential part of the safety verification of this technology, and generating challenging scenarios is critical for scenario-based testing of AVs. Research presented in this dissertation focuses on developing a methodology for crash sequence analysis which is used to generate scenarios for AV safety testing. AVs with SAE Driving Automation Levels 3 and 4 are expected to share the roads and handle conflicts with human drivers. Building a scenario library based on comprehensive samples of historical crash data would be the most efficient way to set up the foundation of a scenario-based AV verification system. Crash scenarios are temporally ordered scenes that consist of 1) participants' actions and interactions, and 2) the relatively static surrounding environment. To incorporate both elements, this dissertation's scenario-generating procedure included two steps - 1) characterization of crashes based on sequences of events, and 2) specification of interrelationships between crash sequences and other crash attributes that depict the surrounding environment. Research tasks developed and demonstrated the crash-sequence-based scenario-generating procedure with three studies.In the first study, a first-of-its-kind crash sequence analysis methodology was developed to serve as the foundation of this dissertation research. In the second study, crash sequence analysis methods were applied to California AV collision data to query and identify representative crash sequence types. In the third study, the scenario-generating procedure incorporates a sequence analysis and a Bayesian network analysis. This dissertation contributes to the understanding of traffic crashes and efficient testing of AVs by developing a first-of-its-kind crash sequence analysis methodology and a novel test scenario generating procedure. This dissertation laid the foundation for traffic crash sequence analysis and the use of crash data for AV test scenario generation. As future crashes happen, new data can be added to the database to add greater depth and further understanding to the critically important topic of scenario-based AV safety evaluation. Findings from this dissertation will have further influences in improving transportation safety and supporting the transition into automated transportation. Knowledge of crash sequences will help future research in analyzing crash causations. A comprehensive test scenario library will speed up large-scale AV safety testing with the help of simulation.

EVIDENCE IN TRAFFIC CRASH INVESTIGATION AND RECONSTRUCTION begins with a detailed description of the entire investigation process. The material then graduates into the various phases and levels of investigations, showing the levels of training and education normally associated with the levels of investigations and consequently the duties and responsibilities of the investigator and reconstructionist. Using narrative, schematics, and photographs, the mechanical inspection process is described in detail by identifying various vehicle parts, explanations of their functions, and methods of identifying failures. Human-related factors in traffic crash investigations are discussed at length, including the traffic crash viewed as a systems failure. Looming vulnerability, a recently developed theoretical construct that helps to describe and understand social, cognitive, organizational, and psychological mechanism, is described. Discussed also is the role of vision in driver performance; perception as a four-way process; perceptions and reactions; driver's reaction to stress; and the roles of pathologists, medical examiners, and coroners in traffic crash reconstruction. Who is an expert and expert evidence are described in detail. Errors that can occur in the investigation process and the tolerances that should be considered or allowed are explained. The manual also discusses the importance of calling upon the skills and advice of occupational specialists, such as reconstructionists, lawyers, traffic engineers, pathologists, medical examiners and others, to assist in the investigation and reconstruction of a crash that will ensure that the objectives of a thorough and complete investigation will be satisfied. Considerable effort has been made in the manual to explain how to identify, interpret and analyze all forms of highway marks and damages that can be used in the reconstruction of a vehicle-related crash. As a guide for investigators, prosecutors and defense attorneys, checkboxes are provided with many of the major topics that can be used as prompters in evaluating the thoroughness of an investigation or for those areas that might or might not need additional coverage at trial or litigation proceedings. To meet international requirements, mathematical references are described in both English (U.S.) and SI (metric) measurement systems, accompanied by various appendices covering symbols and mathematical conversions. Finally, there is a comprehensive quick-find index that takes the reader directly to any topic, formulae, or subject matter - or any combination of these.

Transportation Research Record contains the following papers: Method for identifying factors contributing to driver-injury severity in traffic crashes (Chen, WH and Jovanis, PP); Crash- and injury-outcome multipliers (Kim, K); Guidelines for identification of hazardous highway curves (Persaud, B, Retting, RA and Lyon, C); Tools to identify safety issues for a corridor safety-improvement program (Breyer, JP); Prediction of risk of wet-pavement accidents : fuzzy logic model (Xiao, J, Kulakowski, BT and El-Gindy, M); Analysis of accident-reduction factors on California state highways (Hanley, KE, Gibby, AR and Ferrara, T); Injury effects of rollovers and events sequence in single-vehicle crashes (Krull, KA, Khattack, AJ and Council, FM); Analytical modeling of driver-guidance schemes with flow variability considerations (Kaysi, I and Ail, NH); Evaluating the effectiveness of Norway's speak out! road safety campaign : The logic of causal inference in road safety evaluation studies (Elvik, R); Effect of speed, flow, and geometric characteristics on crash frequency for two-lane highways (Garber, NJ and Ehrhart, AA); Development of a relational accident database management system for Mexican federal roads (Mendoza, A, Uribe, A, Gil, GZ and Mayoral, E); Estimating traffic accident rates while accounting for traffic-volume estimation error : a Gibbs sampling approach (Davis, GA); Accident prediction models with and without trend : application of the generalized estimating equations procedure (Lord, D and Persaud, BN); Examination of methods that adjust observed traffic volumes on a network (Kikuchi, S, Miljkovic, D and van Zuylen, HJ); Day-to-day travel-time trends and travel-time prediction form loop-detector data (Kwon, JK, Coifman, B and Bickel, P); Heuristic vehicle classification using inductive signatures on freeways (Sun, C and Ritchie, SG).

As part of the national effort to improve aviation safety, the Federal Aviation Administration (FAA) chartered the National Research Council to examine and recommend improvements in the aircraft certification process currently used by the FAA, manufacturers, and operators.

Traffic crashes may not always result in severe or fatal injuries, but they can still have nontrivial impacts on system performance, particularly during heavy traffic conditions. One way toward reducing the frequency of such incidents is to first identify the necessary circumstances that resulted in the collision. However, road crashes, particularly intersection related crashes, are complex phenomenon and often result from different combinations of causal factors. Recently, methods for recording high-resolution arterial traffic data have been developed, and it is important for traffic safety engineers to explore such high-resolution data to understand the causes of crashes. In this research one such integrated event based system, known as SMART SIGNAL, which collects and stores detailed loop detector and signal activity, was used to identify the events leading to a crash or a potential crash and illuminate the mechanisms by which traffic conditions and driver decisions interact to produce those events. Two specific event types, a signal violation crash and vehicle pedestrian crash, were evaluated. For the signal violation crash study, SMART SIGNAL data were used to identify the incident and the vehicles involved in the crash. It was then shown how high-resolution data could support a traditional reconstruction of this crash. For vehicle pedestrian interactions, detector and signal activity data were used to predict pedestrian crash risk in the absence of clearance interval at three signalized intersections. A simulation-based method was used to first estimate crash probabilities, and then a counterfactual approach to calculate the probability of the absence of the all-red phase as a necessary condition for the occurrence of the crash provided an alternate estimate of crash-reduction factors for the all-red phase.

The overall objectives of this study were to (1) identify and explore alternative safety data sources and analysis perspectives and (2) demonstrate the potential utility of these alternative approaches in increasing understanding of precipitating events and predisposing factors for crashes occurring on horizontal curves and at unsignalized intersections along rural two-lane roads. Generalized conceptual crash model frameworks were developed, informed by a review of supporting published literature on conceptual crash models and contributing factors, alternative approaches to accident analysis, and the role of constraints in systemic approaches to accident analysis. The frameworks proved useful from several perspectives, including (1) identifying and organizing all factors that influence the likelihood of a crash and defining the event sequences that lead to a crash, (2) providing terminology that will encourage clear communication across accident analysis disciplines as research on crash causation continues, (3) visualizing the nature by which a certain factor influences the likelihood of a crash or by which an event directly causes a crash, and (4) identifying data needs (versus data availability) for studying the precipitating events, system constraints, predisposing factors, and target groups associated with a specific crash type. After marrying the conceptual crash model framework with available data, a study was conducted to determine whether crash causal types, or similar crashes grouped together based on their key precipitating events, could be developed from data, photographs, and narratives developed from detailed, on-scene crash investigations available in the National Motor Vehicle Crash Causation Survey. This was followed by a set of three additional studies primarily focused on alternative safety data sources and analysis perspectives related to predisposing factors. Enhanced data collection and subsequent analysis were demonstrated for three high-priority crash scenarios on rural two-lane roads: “straight crossing path crashes” at unsignalized intersections, combination “control loss/no vehicle action” and “road edge departure/no maneuver” single-vehicle crashes on horizontal curves, and “opposite direction/no maneuver crashes” on horizontal curves. Findings demonstrate that expanding beyond traditional databases used for crash-based evaluations can provide further insight into these crashes. One follow-on analysis in the final part of the study indicated that the alternative approaches to estimating disaggregate measures of exposure, kriging, and quasi-induced demand techniques show some promise and should be considered in future research.