Advancing Automated Mobility: AI-Based Traffic Data Extraction and AI-Powered Modeling of Human–Machine Dynamics

This session presents an integrated research effort that combines large-scale trajectory data collection with AI-enhanced traffic modeling to advance the understanding of Human Driven Vehicle (HDV) and Connected and Automated Vehicle (CAV) interactions. At the core of this work is the Third Generation Simulation Dataset (TGSIM), which introduces real-world vehicle and pedestrian trajectories collected through innovative aerial and infrastructure-based videography. Using advanced computer vision and deep learning methods, high-fidelity trajectories are extracted to capture critical dynamics of car-following, lane-changing, weaving, and pedestrian–vehicle interactions in diverse highway and urban environments. Building on this empirical foundation, the Next Generation Models for CAVs (NGM-CAV) project develops AI-driven behavioral and decision-making models that integrate deep learning, reinforcement learning, and stochastic approaches to represent heterogeneous human responses and automated driving behaviors. These models are calibrated and validated on TGSIM and other large-scale datasets, enabling microsimulation frameworks for Advanced Driver Assistance Systems (ADAS) and Automated Driving Systems (ADS) with unprecedented realism. By unifying AI, behavioral science, and traffic simulation, this framework equips researchers and policymakers with next-generation AI tools to evaluate and optimize the safety and mobility impacts of emerging automated mobility technologies.