DT-IV: From Agent-Level to System-Level Twins is a proposed full-day workshop focused on how digital twins are reshaping intelligent vehicles, connected infrastructure, and city-scale mobility systems. The program is designed to connect high-fidelity vehicle and scene twins with larger operational twins for corridors, cities, and policy-facing transportation platforms.
Tentative full-day format with keynote, technical talks, panel discussion, and cross-scale Q&A.
Morning focus on agent-level twins and afternoon focus on system-level twins.
Designed to bring together university researchers, national labs, mobility testbeds, and industry platform builders.
The workshop bridges digital twins of individual vehicles and local environments with broader twins of transportation systems, urban operations, and infrastructure management.
These short descriptions refine the proposal language using recent speaker, lab, and project materials related to digital twins, physical AI, and transportation systems.
Recent Mcity work emphasizes open-source test-facility twins, remote-access experimentation, and tighter integration between physical proving grounds and virtual validation environments. His keynote is framed around how AV evaluation can move from controlled test tracks toward scalable, naturalistic and eventually city-connected digital twin ecosystems.
The proposal now highlights mixed digital twins for vehicle-road-cloud integration, cooperative control, and modern scenario generation. This better reflects his work spanning mixed-reality validation, multi-vehicle experimentation, and newer generative simulation directions such as TeraSim and TeraSim-World.
The topic description is aligned with the University of Delaware Scaled Smart City line of work, where digital twins support closed-loop algorithm development before physical deployment. The emphasis is on indoor CAV testbeds that reduce the gap between clean simulation and error-prone real-world implementation.
The revised wording stresses high-fidelity visual twins, scenario generation, and sensor-realistic virtual environments. This better fits ongoing Mcity and broader AV research trends around visual world building, sim-to-real transfer, and physically grounded digital replicas for perception evaluation.
The proposal now foregrounds digital-twin-enhanced active safety analysis in mixed traffic. The talk description centers on integrating multi-source roadway data, high-fidelity simulation, and AI-based safety analytics to assess interactions among automated and human-driven vehicles.
The system-level session description is sharpened toward risk-aware urban mobility twins, real-time traffic sensing, and safety-informed city analytics. This reflects both DT-DIMA-style risk-aware mobility analytics and more recent urban digital twin work for emergency-response planning.
The updated text emphasizes scenario-driven digital twins, photo-realistic corridor and city construction pipelines, and anything-in-the-loop evaluation. It also connects ORNL’s CAVE and Real-Sim environments to hardware-linked validation of intelligent mobility strategies.
Focus on world foundation models for physical AI, synthetic data generation, video processing, evaluation, and post-training workflows that support scalable digital twin and simulation ecosystems.
Focus on AI-driven city and infrastructure twins for planning, resilience, and intelligent urban decision support, complementing the physical AI and synthetic data perspective with urban-scale deployment and analytics.
The organizing team combines expertise in transportation systems, digital twins, simulation, AI, and intelligent mobility deployment.
C2SMART Center, New York University · Lead organizer · fan.zuo@nyu.edu
Oak Ridge National Laboratory · Lead organizer · chenh1@ornl.gov
Rochester Institute of Technology · zxbite@rit.edu
C2SMART Center, New York University · kaan.ozbay@nyu.edu