KEYNOTE: Traffic simulation modeling: combining generative modeling with transportation science ...
summary
Extends meso SUMO with link-level fundamental diagrams via a speeds table and reports roughly 10% normalized RMSE improvement in ramp-to-ramp travel times while moving simulations from systematically underestimating congestion to sometimes overbuilding it (rate-of-build remains a tuning issue). Replaces single-point OD calibration with a generative model that only converges after adding a traffic-theory control variate — a variance reduction trick using a simple linear flow metamodel — and shows fitting flow covariances shrinks the continuum of plausible ODs and narrows output uncertainty.
AI-written summary · about
chapters (7)
- 0:00Digital twins in urban mobility
- 1:06Expanding meso SUMO
- 16:08Calibrating demand inputs
- 23:07Reducing OD estimation uncertainty
- 33:40Underdetermination in simulation outputs
- 41:25Open research challenges
- 44:53Closing remarks & Q&A
related talks
13mExtending SUMO for Lane-Free Microscopic Simulation of Connected and Automated Vehicles
20mSimulating platooned connected autonomous vehicle in a large scale urban scenario
17mSimulation based method for the analysis of energy-efficient driving algorithms
15mComputational Mobility: Learning, Prediction, and Algorithms
19mTestbed for Development of AI Solutions to Boost Operational Efficiency in Transportation Networks
18mSUMO Simulations for Federated Learning in Communicating Autonomous Vehicles
20mAI on the Fast Lane: Smart Traffic Systems to Unclog our Cities
16m(Travel) time is money! Costing turns, intersections, and ways