Search

Abstract: The Introduction of Connected-Automated Vehicle (CAV) technology provided a new opportunity to fix the traditional transportation system. Automated vehicles (AuV) would take the driving responsibility and drive the vehicles by analyzing their’ surrounding through a range of sensors. The connectivity feature of these vehicles would facilitate to sense of the roadway and traffic conditions beyond the range of sensors and make informed decisions. While the vehicles equipped with these technologies becoming more common, large-scale market penetration will take a long time. Hence, our transportation infrastructure will pass through a transitional phase where both Human-driven vehicles (HuV) and AuVs share the roadway. Additionally, the prosperity and acceptance of these technologies depend on a clear understanding of the implications of overcoming the limitations of the traditional transportation system. My research focused on developing a comprehensive modeling framework to establish numerical simulation of both types of vehicles (i.e., HuVs, AuVs ) while recognizing the variations of driving behaviors of human drivers. Modeling both vehicle types provided the opportunity to explore diverse mixed traffic scenarios to attain extensive insights into such traffic conditions. Prior to developing the modeling framework, the variations of the human driving patterns were identified through extensive analysis of real-world human driving data. Bi-directional (i.e., longitudinal, lateral) control features were analyzed to comprehend human instincts during driving which can be integrated with the human driver modeling. Further analysis was performed to classify driving behaviors based on these features for the short and long term. The upsides of studying human driving behavior rest not only on better understanding for modeling human drivers but also on designing automated vehicles capable of addressing the variations of human driver behavior. The behavioral classification approach in this part of the research used three vehicular features known as jerk, leading headway, and yaw rate to classify human drivers into two groups (Safe and Hostile Driving) on short-term classification, and drivers’ habits are categorized into three classes (Calm Driver, Rational Driver, and Aggressive Driver). Through the proposed method, behavior classification has been successfully identified in 86.31 ± 9.84% of speeding and 87.92 ± 10.04% of acute acceleration instances. Afterward, the foundation of mixed traffic modeling was developed through car-following strategy formulation. This part of the research proposes a naïve microscopic car-following strategy for a mixed traffic stream in CAV settings and measured shifts in traffic mobility and safety as a result. Additionally, this part of the research explores the influences of platoon properties (i.e. Intra-platoon Headway, Inter-platoon Headway, Maximum Platoon Length) on traffic stream characteristics. Different combinations of HuVs and AuVs are simulated in order to understand the variations of improvements induced by AuVs in a traffic stream. Simulation results reveal that grouping AuVs at the front of the traffic stream to apply CACC-based car-following model will generate maximum mobility benefits for the traffic. Higher mobility improvements can be attained by forming long, closely spaced AuVs at the cost of reduced safety. To achieve balanced mobility and safety advantages from mixed traffic movements, dynamically optimized platoon configurations should be determined at varying traffic conditions and AuVs market penetrations. Finally, grounded on prior research on human driving behavior and modeling framework of mixed traffic, this research objectively experimented with bi-directional motion dynamics in a microscopic modeling framework to measure the mobility and safety implications for mixed traffic movement in a freeway weaving section. This part of research begins by establishing a multilane microscopic model for studied vehicle types from model predictive control with the provision to form a CACC platoon of AuV vehicles. The proposed modeling framework was tested first with HuV only on a two-lane weaving section and validated using standardized macroscopic parameters from the HCM. This model was then applied to incrementally expand the AuV share for varying inflow rates of traffic. Simulation results showed that the maximum flow rate through the weaving section was attained at a 65% AuV share while steadiness in the average speed of traffic was experienced with increasing AuV share. Finally, the results of simulated scenarios were consolidated and scaled to report expected mobility and safety outcomes from the prevailing traffic state as well as the optimal AuV share for the current inflow rate in weaving sections.

Abstract: Variable Speed Limit (VSL), an active traffic demand management (ATDM) strategy, is widely exercised to improved freeway operations by influencing collective driving pattern during congested condition. Since high traffic flow coupled with elevated lane changing rate can onset congestion on freeway bottleneck locations, VSL benefits should be correlated with its’ proficiency in modifying lane changing patterns of traffic stream. Additionally, different compliance level (CL) of VSL should have different impact on congestion reduction and thus on lane changing pattern. Existing macroscopic lane changing models are proved to be inadequate to address the influence of two major aspects of traffic flow variables (speed difference and density difference) on lane changing decisions. Since most of the earlier lane changing models considered either one of them as lane changing incentive, their estimation was found to be impressionable by roadway traffic condition. This study introduces a model framework to approach the problem using simple microscopic lane-changing concept within a macroscopic model. The proposed model considers both speed and density difference as lane-changing incentive based on prevailing roadway condition. The resulting macroscopic model was calibrated and validated with empirical data. The results showed that the proposed model was able to competently reproduce the expected lane changing flows at different traffic state. The proposed lane changing model was applied to estimate lane changing from field observations at a recurrent bottleneck section for ‘without’ and ‘with’ VSL control cases. The primary objective of this research is to study the effect of optimal VSL control on collective lane changing by modeling several compliance levels to VSL and also to study field observations in order to identify the transformation in lane changing behavior due to VSL Control. In this study, several CL to VSL strategies has been modeled to quantify the transformation in lane changing pattern at different compliance level. The model data analysis have shown definite correlations between CL and lane changing flow. Model output also demonstrated that mobility benefits can be achieved through VSL control with improved compliance. To investigate this conjecture, the field test data of Advisory VSL control on Whitemud Drive, Edmonton, Alberta, Canada have been analyzed. The research findings demonstrated that voluntary compliance to VSL has insignificant implications on compliance behavior and consequently collective lane changing pattern. To obtain substantial advantage, the VSL compliance improvement will be necessary which will eventually bring favorable change in collective lane changing pattern.