SIGNATURE WORK
CONFERENCE & EXHIBITION 2024

In my signature work thesis, I delve into the realm of Traffic Signal Control (TSC) methods with the objective of mitigating travel congestion in urban environments. The journey begins with a comprehensive survey encompassing both traditional Rule-based systems and cutting-edge Reinforcement Learning (RL) models. Through this survey, I analyse the fundamental principles, evaluate performance metrics, and scrutinize the efficacy of each approach in managing traffic flow at intersections.

Transitioning from theoretical inquiry to practical implementation, empirical assessments are conducted using the CityFlow simulator, complemented by diverse datasets. These empirical analyses serve as the crucible for evaluating the performance of TSC models under standardized conditions.

Central to my findings is the recognition of the pivotal role played by dataset complexity and hyper-parameter optimization in shaping the efficacy of TSC methods. I underscore the significance of these factors, especially in the context of complex urban environments where traffic patterns exhibit variability and unpredictability. By showcasing the interplay between dataset characteristics and model performance, my research guides toward the design and implementation of more adaptive and resilient TSC strategies. Hence, my work represents a contribution to the field of traffic engineering, offering actionable insights into the optimization of TSC methods.