Opening Remarks and Workshop Goals
Organizers introduce the DT-IV scope, the agent-to-system framing, and the central question of how digital twins can support trustworthy intelligent-vehicle development and transportation operations.
Tentative full-day program
Event Schedule / Details
Times presume a 09:00–17:00 schedule. Speaker order and confirmation status may change. The current structure is designed to move from agent-level twins and testbeds toward city-scale twins, physical AI platforms, and deployment questions.
From Test Tracks to City Digital Twins: Validating Automated Vehicles in Naturalistic Environments
Henry X. Liu, University of Michigan / Mcity. Proposed keynote on how open, test-facility twins and remote-access validation environments can reduce the gap between controlled experimentation and real-world deployment.
Mixed Digital Twins for Vehicle-Road-Cloud Integration and Cooperative Control
Jiawei Wang, University of Michigan. Focus on mixed digital twin architectures, cooperative driving validation, and scenario generation workflows that connect physical assets, simulations, and connected infrastructure.
Indoor CAV Testbeds and 3D Twins for Edge and Onboard Vehicle Computing
William He, University of Delaware. Discussion of indoor and scaled-city experimentation, algorithm validation under realistic communication and sensing imperfections, and how 3D twins help move ideas from simulation to testbed deployment.
Refreshment Break
Informal networking and transition to visual twin and perception-focused talks.
High-Fidelity Visual Digital Twins for AV Perception and Scenario Generation
Minghan Zhu / Yue Hu, University of Michigan / Mcity. Expected focus on visually rich digital replicas, sensor-realistic simulation, and scenario generation pipelines for perception and closed-loop AV testing.
AI-Empowered Digital Twins for Active Safety in Mixed Traffic
Yang Zhou, Texas A&M University. Session on digital twin-based active safety analysis, high-fidelity traffic environments, and AI-assisted evaluation of interactions among automated and human-driven vehicles.
Agent-Level Q&A and Mini-Panel
Moderated discussion on data requirements, validation standards, sensor realism, and reproducible design patterns for local and vehicle-level twins.
Lunch Break
Open networking period.
NVIDIA Cosmos for Physical AI and World-Scale Transportation Scenarios
NVIDIA Cosmos Team. Revised title to foreground world foundation models, controllable synthetic data, video evaluation, and post-training workflows relevant to autonomous vehicles and physically grounded simulation.
Driver Risk-Aware Urban Mobility Twins: From Trajectories to City-Scale Analytics
Zilin Bian, Rochester Institute of Technology. Session focused on urban digital twins that combine traffic sensing, mobility analytics, and safety risk prediction to support proactive operations and planning.
System-Level Q&A I
Discussion on how system-scale twins ingest data from agent-level models, sensors, and testbeds, and how they support monitoring, control, and scenario-based policy analysis.
Refreshment Break
Transition to the final system-level talks and closing discussion.
RealTwin: Scenario-Driven Digital Twins for AV Safety and Operations
ORNL RealTwin Team, Oak Ridge National Laboratory. Focus on scenario libraries, digital twin city construction, hardware-linked evaluation, and integrated virtual-physical testing environments for intelligent mobility research.
AI-Driven City and Infrastructure Twins for Planning and Resilience
Tianming Liu, University of Michigan. Session centered on AI-enabled modeling of urban systems and infrastructure, with emphasis on scalable planning support, intelligent urban analysis, and the broader city-scale twin agenda.
System-Level Q&A II
Moderated closing discussion on standards, data governance, interoperability, and how agent-level and system-level twins can be coupled into a more robust research and deployment roadmap.
Industry interest listed in the proposal includes NVIDIA, Autoware, and Wayve. Final invitations, titles, and speaker ordering can be updated as the workshop lineup evolves.