Reconstruction of Mixed Traffic Systems at Micro and Macro Scales [ITS Ph.D.s]

Understanding complex traffic dynamics, characterized by the interplay between vehicle automation technologies, human drivers, and diverse transportation modes, presents significant challenges for transportation research. The integration of autonomous vehicles with human-driven vehicles introduces new dynamics into traffic flow, necessitating a comprehensive understanding of their coexistence. Additionally, the heterogeneity of various vehicle classes further complicate traffic patterns. Advancements in traffic data collection, particularly camera-based trajectory data, provide a wealth of information for analyzing microscopic details and macroscopic trends in traffic behavior. However, challenges in modeling and estimation techniques, such as accuracy, robustness and scalability need to be addressed. Furthermore, data quality issues need to be tackled in a systematic manner.

This dissertation addresses the overarching question of developing comprehensive and efficient methods to understand complex traffic dynamics. It solves a series of reconstruction problems, involving parameter estimation, state estimation, and data reconciliation. Parameter estimation focuses on inferring unknown parameters of microscopic traffic models from on-board vehicle sensor data. State estimation involves estimating current or past states of macroscopic traffic dynamics using fixed-location measurements. Data reconciliation processes incomplete and noisy video tracking data while adhering to dynamical constraints. The integration of these tools provides insights into complex traffic patterns at various scales.