Perceptions of vulnerable roadway users on autonomous vehicle regulations.

INTRODUCTION: Development and implementation of autonomous vehicle (AV) related regulations are necessary to ensure safe AV deployment and wide acceptance among all roadway users. Assessment of vulnerable roadway users' perceptions on AV regulations could inform policymakers the development of appropriate AV regulations that facilitate the safety of diverse users in a multimodal transportation system. METHOD: This research evaluated pedestrians' and bicyclists' perceptions on six AV regulations (i.e., capping AV speed limit, operating AV in manual mode in the sensitive areas, having both pilot and co-pilot while operating AVs, and three data-sharing regulations). In addition, pedestrians' and bicyclists' perceptions of testing AVs in public streets were evaluated. Statistical testing and modeling techniques were applied to accomplish the research objectives. RESULTS: Compared to the other AV regulations assessed in this research, strong support for AV-related data sharing regulations was identified. Older respondents showed higher approval of AV testing on public roadways and less support for regulating AVs. AV technology familiarity and safe road sharing perceptions with AVs resulted in lower support for AV regulations. CONCLUSIONS: Policymakers and AV technology developers could develop effective educational tools/resources to inform pedestrians and bicyclists about AV technology reliability and soften their stance, especially on AV regulations, which could delay technology development. PRACTICAL APPLICATIONS: The findings of this research could be used to develop informed AV regulations and develop policies that could improve pedestrians' and bicyclists' attitudes/perceptions on regulating AVs and promoting AV technology deployments.

Identification of high-risk roadway segments for wrong-way driving crash using rare event modeling and data augmentation techniques.

Wrong-Way Driving (WWD) crashes are relatively rare but more likely to produce fatalities and severe injuries than other crashes. WWD crash segment prediction task is challenging due to its rare nature, and very few roadway segments experience WWD events. WWD crashes involve complex interactions among roadway geometry, vehicle, environment, and drivers, and the effect of these complex interactions is not always observable and measurable. This study applied two advanced Machine Learning (ML) models to overcome the imbalanced dataset problem and identified local and global factors contributing to WWD crash segments. Five years (2015-2019) of WWD crash data from Florida state were used in this study for WWD model development. The first modeling approach applied four different hybrid data augmentation techniques to the training dataset before applying the XGBoost classification algorithm. In the second model, a rare event modeling approach using the Autoencoder-based anomaly detection method was applied to the original data to identify WWD roadway segments. A third model was applied based on the statistical method to compare the performance of ML models in predicting the WWD segments. The performance comparison of the adopted models showed that the XGBoost model with the Adaptive Synthetic Sampling (ADASYN) method performed best in terms of precision and recall values compared to the autoencoder-based anomaly detection method. The best-performing model was used for the feature analysis with an interpretable machine-learning technique. The SHapley Additive exPlanations (SHAP) values showed that high-intensity developed land use, length of roadway, log of Annual Average Daily traffic (AADT), and lane width were positively associated with WWD roadway segments. The results of this study can be used to deploy WWD countermeasures effectively.

Application of Bayesian Space-Time interaction models for Deer-Vehicle crash hotspot identification.

The objective of this research was to identify and prioritize deer-vehicle crash (DVC) hotspots using five years of crash data. This study applied Bayesian spatiotemporal models for the identification of the DVC hotspots. The Bayesian spatiotemporal model allows to observe area-specific trends in the DVC data and highlights specific locations where DVC occurrence is deteriorating or improving over time. Census Tracts (CTs) were used as the geographic units to aggregate DVC, land use, and transportation infrastructure related data of Minnesota (MN) for the year 2015 to 2019. Several tests were conducted to evaluate the performance of the hotspot identification methods. The result showed that Type-I spatiotemporal interaction model (Model-2) outperforms other four space-time models in terms of predicting DVC frequency in CTs and hotspot identification performance test measures. Results showed that forest area, vegetation, and wetland percentages were positively associated with DVC frequency, whereas the percentage of developed land use was negatively associated with DVC frequency. The findings of this study suggest that the deer population plays an important role in DVCs, which indicates that deer population management is necessary to minimize the DVC risks. Using the final Type-I spatiotemporal interaction model, 65 "High-High" CTs were identified, where both the posterior mean of the decision parameter (potential for safety improvement) and the area-specific trend were higher. The distribution of the identified hotspots showed that the risk of DVCs was more in suburban areas with mixed land use conditions. These CTs represent high-risk zones, which need immediate safety improvement measures to reduce the DVC risks. As DVC can occur at any roadway segment location, DVC hotspots information is important for safety engineers and policymakers to implement area specific countermeasures.