Data-driven solutions to intelligent transportation system problems concerning traffic safety and mobility
Description
TitleData-driven solutions to intelligent transportation system problems concerning traffic safety and mobility
Date Created2021
Other Date2021-05 (degree)
Extent1 online resource (x, 128 pages)
DescriptionIssues surrounding traffic safety and mobility have been at the center of transportation research for many years. Much effort has been made to identify the problematic road features and develop countermeasures to mitigate the crash risk in crash-prone locations. With the emerging connected vehicle technology and smart roadways, the need for smart adaptive traffic signal control and Urban Navigation is more than ever. This dissertation starts with a preliminary study to show how traffic signal timing and drivers’ behavior impact mobility and safety measures at a network of two intersections and the connecting roadway segments. This study integrates the mobility and safety measures into a single two-dimensional Key Performance Indicator (KPI) for intersections. A SUMO simulation model of a real-world roadway network in China is built and run for many scenarios based on a design of experiment aiming at identifying a statistical correlation between the combined measures, the timing of the traffic signals, and drivers’ behaviors. In the world of smart roadways and vehicles, the findings from this work can be used to alert drivers adequately and to optimize signal timings at intersections through closer monitoring of roadway traffic and drivers’ behavior. This model is also appropriate for road safety audits (RSAs) to prioritize improvements with respect to the combined key performance indicator.
The work continues by proposing a novel approach, the Safe Route Mapping (SRM) model that integrates crash-based estimates with conflict risks computed from driver-based data to score the safety of roadways. An advanced Safety Performance Function (SPF) estimates the number of crashes, and a driver-based model computes dynamic conflict risk measures from driver and traffic data. In real-life implementations of the proposed methodology, the driver-based data and traffic data can be collected from vehicles or infrastructure-based data sources, including smartphones. The methodology uses real historical crash data and simulated driver-based data obtained from VISSIM and SSAM to show safety risk heat maps of a roadway and illustrate how these maps change with driver types and traffic volumes. The proposed methodology fills the existing gaps in near real-time dynamic data to designate safe corridors, dispatch law enforcement, and plan safety projects. Drivers can also use road heat maps for situational awareness and trip planning.
Followed by the safety study, an Accumulated Exponentially Weighted Waiting Time-based Adaptive Traffic Signal Control (AEWWT-ATSC) model is proposed to calculate roadways' priorities for signal scheduling. As the size of the traffic network grows, it adds significant complexity and challenges to computational efficiencies. A novel Distributed Multi-agent Reinforcement Learning (DMARL) with a graph decomposition approach for large-scale ATSC problems is proposed to solve this issue. The method clusters intersections by the level of connectivity (LoC), defined by the average residual capacities (ARC) between connected intersections, making it possible to train subgraphs instead of the entire network in a synchronized way. The problem is formulated as a Markov Decision Process (MDP), and the Double Dueling Deep Q Network with Prioritized Experience Replay is utilized to solve it. Under the optimal policy, the agents can select the optimal signal time to minimize the waiting time and queue size. The evaluation shows that the superiority of the AEWWT-ATSC based RL methods in different densities demonstrates the DMARL with graph decomposition approach on a large graph in Manhattan, NYC. The approach is generic and can be extended to various types of use cases.
In the last part of the work, a data-driven optimization approach for dynamic shortest path problems (DDSP) considering traffic safety for urban navigations is developed. The dynamic risk scores and travel times on different road facilities at different times and locations are estimated by the Safe Route Mapping (SRM) methodology and Long Short Term Memory (LSTM) with Autoencoder, respectively, where possible variations in the future are considered. The DDSP is formulated as a mixed-integer linear programming problem under risk constraints to minimize the total travel cost, defined as the weighted sum of distance and travel time. An improved Double Search Algorithm (DSA) with alternative initial-solution algorithms is designed to accommodate various problem scales and improve the efficiency of the DDSP. Moreover, the subgraph and self-adaptive insertion are adopted as acceleration strategies to improve computational efficiency further. Numerical experiments investigate the algorithm's computational performance and compete with the CPLEX solver, a label-setting algorithm, a state-of-the-art algorithm, and commercial navigation software. The result shows a satisfactory trade-off between optimality and computational efficiency with the proposed acceleration strategies. The real-life implementation shows that the algorithm can provide the same quality of routing decisions on the shortest and fastest path as the Google Map, which is promising for Urban Navigation.
NotePh.D.
NoteIncludes bibliographical references
Genretheses, ETD doctoral
LanguageEnglish
CollectionSchool of Graduate Studies Electronic Theses and Dissertations
Organization NameRutgers, The State University of New Jersey
RightsThe author owns the copyright to this work.
Statistical Profile