<p>In the context of highway safety factors, road geometrics and pavement condition are of particular interest to highway managers as they fall within their direct control and therefore can be addressed through highway projects. In spite of the preponderance of econometric modeling in highway safety research, there still remain areas worthy of further investigation. These include 1) the lack of sufficient feedback to roadway preservation engineers regarding the impacts of road-surface condition on safety; 2) the inadequate feedback to roadway designers on optimal lane and shoulder width allocation; 3) the need for higher predictive capability and reliability of models that analyze roadway operations; and 4) the lack of realistic simulations to facilitate reliable safety impact studies regarding autonomous vehicles (AV). In an attempt to contribute to the existing knowledge in this domain and to throw more light on these issues, this dissertation proposes a novel framework for enhanced prediction of highway safety that incorporates machine learning and econometrics with optimization to evaluate and quantify the impacts of safety factors. In the traditional highway operating environment, the proposed framework is expected to help agencies improve their safety analysis. Using an Indiana crash dataset, this dissertation implements the framework, thereby 1) estimating the safety impacts of the road-surface condition with advanced econometric specifications, 2) optimizing space resource allocations across highway cross-sectional elements, and 3) predicting the fatality status of highway segments using machine learning algorithms. In addition, this dissertation discusses the opportunities and the expected safety impacts and benefits of AV in the emerging operating environment. The dissertation also presents a proposed deep learning-based autonomous driving simulation framework that addresses the limitations of AV testing and evaluation on in-service roads and test tracks.</p>

Roadway safety continues to be a priority for the Utah Department of Transportation (UDOT) Traffic and Safety Division. UDOT has participated in and managed several research projects in recent years to determine the roadway segments of highest safety concern in the state. This research has provided UDOT with more tools to assist in safety project prioritization. Researchers in Department of Civil and Environmental Engineering at Brigham Young University (BYU) have worked with UDOT and the Statistics Department at BYU to create two network screening statistical tools called the Utah Crash Prediction Model (UCPM) and the Utah Crash Severity Model (UCSM) to analyze roadway segment safety. The Roadway Safety Analysis Methodology (RSAM) was developed as a process to run these segment models. Because a significant portion of crashes occur at intersections, there is a need to analyze roadway safety specifically at intersections. This research focuses on the development of the Utah Intersection Crash Prediction Model (UICPM) and the Intersection Safety Analysis Methodology (ISAM). The UICPM is a Bayesian generalized linear model that determines crash distributions for each intersection based on roadway characteristics and historical crash data. The observed number of crashes at each intersection is compared with the crash distribution, and a percentile value is calculated as the probability that the number of crashes occurring at an intersection in a particular year is less than or equal to the average annual number of crashes. A high percentile value indicates that more crashes were observed than expected and the intersection is a hot spot and should be considered for safety improvements. All intersections are ranked at the state, UDOT Region, and county levels based on the percentile value, the higher ranks having higher percentile values. The ISAM is the three-step process that was developed to execute the UICPM. The first step is to prepare the model input by formatting and combining the roadway characteristics and crash data files. Crashes are assigned to intersections if they fall with the functional area of an intersection. Due to data limitations, the ISAM is currently being used only for intersections of at least two state routes. It is anticipated that, as more data are made available, the ISAM will function properly for intersections of non-state routes as well. The second step is to execute the UICPM using the R GUI tool and R software. The third step is to create a two-page Intersection Safety Analysis Report (ISAR) for intersections of interest and maps of the state, UDOT Regions, and counties with the model results. Parts of the ISARs are auto-generated and the rest is entered manually by an analyst. The two-page ISARs will be used by UDOT Regions to prioritize intersection safety projects in their respective areas.

The Utah Department of Transportation (UDOT) Traffic and Safety Division continues to advance the safety of the state roadway network through network screening and decision making tools. In an effort to aid UDOT in meeting this goal, the Department of Civil and Environmental Engineering at Brigham Young University (BYU) has worked with the Statistics Department in developing analysis tools for highway safety, specifically the Utah Crash Prediction Model (UCPM) and the Utah Crash Severity Model (UCSM). Additional tools and methodologies, such as the "Hot Spot Identification and Analysis," have been created to summarize the roadway characteristics, crash data, and possible countermeasures of roadway segments with safety problems.This research focuses on the creation of a three part "Roadway Safety Analysis" methodology, which applies and automates the cumulative work of recently completed highway safety research conducted for UDOT. The first part is to prepare the roadway data and crash data for the statistical analysis. The second part is to perform the network screening statistical analysis; rank the segments by state, UDOT Region, and county; and select segments of interest. The third part is to compile and publish the Roadway Safety Analysis reports for the selected segments of interest. These parts are accomplished using the automation tools and graphical user interfaces (GUIs), which are documented in three respective volumes of user manuals. The automation tools and GUIs were developed with checks and processes to allow the Roadway Safety Analysis methodology to be completed with new, updated roadway and crash datasets.The Roadway Safety Analysis methodology allows future iterations of the UCPM and UCSM analysis and compilation of the Roadway Safety Analysis reports to be conducted in a user friendly environment. A series of critical data columns were identified to communicate the need for data consistency for future iterations of this safety research. An example of the entire process of the Roadway Safety Analysis methodology is given to illustrate how the three parts tie together. The overall process has automated data processing tasks, which saves time and resources for the analyst to investigate possible safety measures for segments of interest. Recommendations for future highway safety research are given, including continued development of the Roadway Safety Analysis methodology, an analysis of intersections and horizontal curves, the implementation of the Roadway Safety Analysis methodology to other states, and the advancement of safety countermeasures and geospatial tools for highway safety research.

Tese de Mestrado Integrado. Engenharia Civil (Vias de Comunicação). Faculdade de Engenharia. Universidade do Porto. 2013

This dissertation addresses the relationship between roadway segment length and roadway attributes and their relationship to the efficacy of Safety Performance Function (SPF) models. This research focuses on three aspects of segmentation: segment length, roadway attributes, and combinations of the two. First, it is shown that choice of average roadway segment length can result in markedly different priority lists. This leads to an investigation of the effect of segment length on the development of SPFs and identifies average lengths that produce the best-fitting SPF. Secondly, roadway attributes are filtered to test the effect that homogeneity has on SPF development. Lastly, a combination of segment length and attributes are examined in the same context. In the process of conducting this research a tool was developed that provides objective goodness-of-fit measures as well as visual depictions of the model. This information can be used to avoid things like omitted variable bias by allowing the user to include other variables or filter the database. This dissertation also discusses and offers examples of ways to improve the models by employing alternate model forms. This research revealed that SPF development is sensitive to a variety of factors related to segment length and attributes. It is clear that strict base condition filters based on the most predominant roadway attributes provide the best models. The preferred functional form was shown to be dependent on the segmentation approach (fixed versus variable length). Overall, an important step in SPF development process is evaluation and comparison to determine the ideal length and attributes for the network being analyzed (about 2 miles or 3.2 km for Kentucky parkways). As such, a framework is provided to help safety professionals employ the findings from this research.

Developing sound or reliable statistical models for analyzing vehicle crashes is very important in highway safety studies. A difficulty arises when crash data exhibit overdispersion. Over-dispersion caused by unobserved heterogeneity is a serious problem and has been addressed in a variety ways within the negative binomial (NB) modeling framework. However, the true factors that affect heterogeneity are often unknown to researchers, and failure to accommodate such heterogeneity in the model can undermine the validity of the empirical results. Given the limitations of the NB regression model for addressing over-dispersion of crash data due to heterogeneity, this research examined an alternative model formulation that could be used for capturing heterogeneity through the use of finite mixture regression models. A Finite mixture of Poisson or NB regression models is especially useful when the count data were generated from a heterogeneous population. To evaluate these models, Poisson and NB mixture models were estimated using both simulated and empirical crash datasets, and the results were compared to those from a single NB regression model. For model parameter estimation, a Bayesian approach was adopted, since it provides much richer inference than the maximum likelihood approach. Using simulated datasets, it was shown that the single NB model is biased if the underlying cause of heterogeneity is due to the existence of multiple counting processes. The implications could be poor prediction performance and poor interpretation. Using two empirical datasets, the results demonstrated that a two-component finite mixture of NB regression models (FMNB-2) was quite enough to characterize the uncertainty about the crash occurrence, and it provided more opportunities for interpretation of the dataset which are not available from the standard NB model. Based on the models from the empirical dataset (i.e., FMNB-2 and NB models), their relative performances were also examined in terms of hotspot identification and accident modification factors. Finally, using a simulation study, bias properties of the posterior summary statistics for dispersion parameters in FMNB-2 model were characterized, and the guidelines on the choice of priors and the summary statistics to use were presented for different sample sizes and sample-mean values.

Work zones are visually and physically complex environments, requiring that drivers maintain control of their vehicle and comprehend atypical and often discontinuous traffic control devices to safely navigate appropriate paths. Freeway diverges represent particularly difficult work zones areas. This thesis investigates current methods of delineating diverges in freeway work zones to determine important characteristics of these methods for future research. A virtual environment was constructed with two ramp geometries: a freeway continuing straight and one curving left. Still images of work zones on these geometries were created using drums spaced 10 ft apart, drums spaced 40 ft apart, drums spaced 40 ± 2 ft apart, and portable concrete barriers. These alternatives were used to construct temporary ramps that were either open or closed. 39 participants were asked to identify whether the ramp was open or closed and their responses were recorded to evaluate the performance of each alternative. Results indicate the importance of the Gestalt principles of closure, proximity, and continuity in perception of temporary exit ramps in work zones. These results will be used to guide future research into methods of delineating diverges in freeway work zones.

Safety reviews of existing roads are becoming a popular practice of many agencies nationally and internationally. Knowing road safety information is of great importance to both policymakers in addressing safety concerns and travelers in managing their trips. There have been various efforts in developing methodologies to measure and assess road safety in an effective manner. However, the existing research and practices are still constrained by their subjective and reactive nature. The goal of this research is to develop a framework of Road Risk Indices (RRIs) to assess road risks of existing highway infrastructure for both road users and agencies based on road geometrics, traffic conditions, and historical crash data. The proposed RRIs are intended to give a comprehensive and objective view of road safety, so that safety problems can be identified at an early stage before they rise in the form of accidents. A methodological framework of formulating RRIs that integrates results from crash prediction models and historical crash data is proposed, and Linear Referencing tools in the ArcGIS software are used to develop digital maps to publish estimated RRIs. These maps provide basic Geographic Information System (GIS) functions, including viewing and querying RRIs, and performing spatial analysis tasks. A semi-parameter count model and quantile regression based estimation are proposed to capture the specific characteristics of crash data and provide more robust and accurate predictions on crash counts. Crash data collected on Interstate Highways in Washington State for the year 2002 was extracted from the Highway Safety Information System (HSIS) and used for the case study. The results from the case study show that the proposed framework is capable of capturing statistical correlations between traffic crashes and influencing factors, leading to the effective integration of safety information in composite indices. / text

There have been many documents published that set strategic goals for the future, including transportation-related goals. However, few documents focus heavily on a specific approach to improve transportation safety. Therefore, a supporting policy document focused on transportation safety is required to ensure that the transportation system runs safely and efficiently; a Strategic Highway Safety Plan (SHSP) serves as that document. A SHSP is a high-level traffic safety policy that represents a scientific, data-driven, four to five year comprehensive safety document that is designed to identify a jurisdiction’s emphasis areas (i.e., key areas of safety concern) and target safety goals [i.e., collision reduction goal(s)], and may also include network screening (i.e., identification of high collision locations) and safety strategies/programs for each chosen emphasis area. There are, however, limited documents that discuss the procedure for the development of a SHSP specifically for a municipality. Therefore, the goal of this research was to improve traffic safety by reducing the number and severity of collisions in municipalities across Canada. The objective for this research was to develop a data-driven and more scientific municipal-level SHSP development process (i.e., procedure and key components) that may be used to improve traffic safety for municipalities across Canada. Existing procedures, key components and approaches to develop the key components in existing SHSPs published mainly in North America were reviewed. The literature review (FHWA, 2006; CCMTA, 2011b) suggested that the typical procedure for the development of a SHSP is identifying a “champion” (i.e., an individual or unit with high-level leadership), developing a vision, identifying key stakeholders, developing the key components (i.e., selecting the key emphasis areas, establishing target safety goals, selecting the strategies/programs for the chosen key emphasis areas), and updating and evaluating the SHSP. The existing procedures and key components were adjusted to create the modified process. The modified process consisted of two additional steps to the procedure: 1) Incorporating Upper-Level Policies and 2) Conducting Network Screening. The modified process also outlined the most appropriate approaches to use to develop the key components of a municipal-level SHSP. The modified process (i.e., procedure and key components) was applied to develop a municipal-level SHSP for the City of Saskatoon through a case study to compare the results to the existing process. Saskatoon’s SHSP included seven emphasis areas for a definite period of time (i.e., for the next five years). Target safety goals, network screening and strategies/programs were also developed, but only for the selected emphasis areas. Recent ten-year (2001-2010) collision data from the SGI was used to select emphasis areas, develop target safety goals and conduct network screening. Based on the case study results, upper-level policies should be incorporated in the development of the key components of a municipal-level SHSP. This is because a municipal-level SHSP is the lowest-level SHSP and should incorporate the emphasis areas, target safety goals and strategies/programs that are included in upper-level SHSPs (i.e., provincial- and federal-level). In addition, the SHSP can act as an operational-level safety action plan that supports a jurisdiction’s Strategic Plan. The addition of network screening also provides useful locations in a municipal-level SHSP. The case study results showed that the modified process provided detailed information required by a municipality to make informed safety investment decisions compared to the basic information the existing process provided. Therefore, the modified process is a data-driven and more scientific process that can be used to develop SHSPs that will improve traffic safety for municipalities across Canada.

Dissertação (mestrado)—Universidade de Brasília, Faculdade de Tecnologia, Departamento de Engenharia Civil e Ambiental, 2016. / Submitted by Albânia Cézar de Melo (albania@bce.unb.br) on 2016-05-02T15:22:29Z No. of bitstreams: 1 2016_DanieleRoewerLagemannSWaihrich.pdf: 22832959 bytes, checksum: 5f8ac52b5af727175e96735f4cc7385d (MD5) / Approved for entry into archive by Raquel Viana(raquelviana@bce.unb.br) on 2016-05-03T21:27:19Z (GMT) No. of bitstreams: 1 2016_DanieleRoewerLagemannSWaihrich.pdf: 22832959 bytes, checksum: 5f8ac52b5af727175e96735f4cc7385d (MD5) / Made available in DSpace on 2016-05-03T21:27:19Z (GMT). No. of bitstreams: 1 2016_DanieleRoewerLagemannSWaihrich.pdf: 22832959 bytes, checksum: 5f8ac52b5af727175e96735f4cc7385d (MD5) / Na promoção da segurança é importante o estabelecimento de recursos para uma avaliação quantitativa da segurança no ambiente viário. Os modelos preditivos de acidentes, a partir de técnicas estatísticas adequadas, estimam o número esperado de acidentes em diferentes momentos de um empreendimento viário, podendo, em especial, atuar preventivamente na segurança. No Brasil, o desenvolvimento de modelos preditivos de acidentes não está ainda bem instituído, havendo uma carência de modelos que quantifiquem a segurança em rodovias brasileiras. O manual americano Highway Safety Manual (HSM) apresenta um método preditivo para diferentes configurações viárias e inclui um procedimento de calibração do método, o que possibilita a sua transferência para outras localidades. Nesta dissertação foi realizada a calibração do método preditivo do HSM em rodovias de pista dupla nas regiões de Minas Gerais e Goiás/ Distrito Federal, tendo por resultado um Fator de Calibração para cada uma das regiões estudadas. Na avaliação da transferibilidade do modelo calibrado foram aplicadas medidas de qualidade de ajuste. Os resultados obtidos não confirmam a transferibilidade do modelo original do HSM calibrado nos cenários estudados, pois, embora tenham sido obtidos resultados satisfatórios na região MG, foram também observados resultados precários das medidas de qualidade de ajuste na região GO/DF. Dessa forma, neste estudo inicial em rodovias rurais de pista dupla no Brasil, a transferência do modelo HSM não se confirmou como uma alternativa efetiva, o que não descarta a possibilidade da obtenção de um modelo calibrado com algum grau de sucesso em outras rodovias e para outras regiões. Os retornos e faixas adicionais são dispositivos frequentes nos segmentos rodoviários estudados. Em função disso, foram delimitadas amostras alternativas que buscaram avaliar o impacto destes dispositivos na segurança, uma vez que os mesmos não são incluídos no modelo preditivo original do HSM. A partir desta análise, foram obtidos, apenas em nível exploratório, “Fatores de Modificação de Acidentes” específicos para estes dispositivos. _______________________________________________________________________________________________ ABSTRACT / When promoting safety, it is important to establish resources to carry out a quantitative assessment of safety in the road environment. Accident predictive models, based on proper statistic technics, estimate the expected number of accidents in different moments of a road system, making it possible to act preventively in regards to safety. In Brazil, the development of accident predictive models has not been well established yet. We lack models that quantify safety in Brazilian highways. The American Highway Safety Manual – HSM – presents a predictive method for different road configurations and includes a procedure for method calibration, which enables its application in other parts of the world. In this dissertation, we have carried out the calibration of the HSM predictive method to multilane highways in the Minas Gerais and Goiás/Federal Ditrict regions, having a calibration factor for each studied region as a result. When validating the capacity of the model of being transferred, we have applied measures to assess the adjustment quality of the calibrated model. The results obtained did not confirm the transferability of the original HSM model, which was calibrated according to the studied scenarios. This occurred because, although we have obtained satisfactory results in the Minas Gerais region, we have observed disappointing results regarding the adjustment quality measures of the calibrated model in the GO/DF region. Thus, this initial study on rural multilane highways in Brazil did not confirm the transference of HSM model as an effective alternative. However, this does not rule out the possibility of obtaining a calibrated model with some degree of success in other highways and other regions. U-turns and Passing Lanes are common features in the studied highway segments. For that reason, we have selected alternative samples for the purpose of assessing the impact they had in safety, since they are not present in the original HSM predictive model. Following this analysis, we have obtained, only at an exploratory level, “accident modification factors” specific for such features.