ANALYSIS OF CONNECTED AND AUTOMATED VEHICLES TO ADVANCE MULTIMODAL SAFETY

Transportation Engineers and Planners are working towards improving transportation safety by researching Connected and Automated vehicles (CAV), which are anticipated to address human error factors in crash studies. This dissertation investigates various safety features of CAVs such as Enhanced Pedestrian in Crossing Warning (EPCW), Enhanced Vehicle Turning Right in Front of Bus Warning (EVTRW), Advanced Driver-Assistance Systems (ADAS), and Autopilot feature to understand their performance at this early stage of deployment which can be insightful for CAV researchers. The two research questions addressed in this dissertation are 1) how can the data from CAVs be utilized to increase transit, pedestrian, motor vehicle, and infrastructure safety; and 2) how can the data from CAVs be utilized to understand their readiness and effectiveness. Due to the limitation of publicly available CAV data, knowledge about their safety performances is limited. This dissertation attempts to address a segment of this gap. First, the safety features of connected buses are studied to determine the association between pedestrian and bus-based alerts with crashes that can aid the safety of vulnerable road users (VRUs), transit buses, and intersections. A literature review of Automated shuttles follows this to identify the potential locations of AV shuttle deployment at an early stage. Later, the safety features of AVs are studied by comparing lower-level automated vehicles to understand which AVs perform better in terms of safety at different road types, weather and lighting conditions, and driver behaviors. Various data science, spatial, and frequentist techniques including panel poisson regression model to address heterogeneity, ordered logistic regression model, binary logistic regression model, text analysis, hazardous location identification method in GIS, and novel safety surrogate measures (SSM), are utilized to conduct the analysis. Finally, research implications and future works are discussed to understand these emerging vehicles comprehensively. Long term scientific value of this dissertation is that it harnesses big data from cutting-edge intelligent transportation systems and infrastructures and integrates them with conventional crash data to perform analysis that can aid in advancing multimodal safety. The dissertation will be helpful for city planners, policymakers, civil engineers, designers, and researchers working towards achieving Vision Zero goals.