AI-Enhanced Traffic Safety Digital Twin

A research platform connecting road-user sensing, SUMO simulation, safety analytics, comparative visualization, and a natural-language interface.

University of Washington · Current research

Project story

From question to working system.

This project develops a 2D traffic-safety digital twin for exploring how observed road-user movements and simulated scenarios can support proactive safety decisions.

The platform brings together detection and trajectory tracking, microscopic simulation, surrogate safety measures, composite indices, comparison views, and an LLM-enabled dashboard. It is designed as a transparent research demonstrator in which assumptions and outputs can be inspected rather than hidden.

2D twin

Observed trajectories connected to a microscopic simulation

SSMs

Conflict indicators and composite safety indices

Scenario lab

Side-by-side what-if testing and visualization

Conversational

An LLM interface for transparent exploration

My contributions

Roles and technical work.

System architecture and research design
Trajectory and simulation integration
Safety indicators and composite indices
Comparative visualization
LLM-enabled dashboard design

Project gallery

Systems, fieldwork, and results.

AI-enhanced traffic-safety digital twin presentation — Presenting the digital-twin framework at ASCE ICTD 2026.

Traffic-safety digital twin research exchange — Discussing intelligent sensing, simulation, and decision support.

ASCE ICTD digital-twin session — Conference discussion around the working research demonstrator.

Resources

Presentations, publications, reports, and links.

Public materials for this project are being prepared. Request project materials ↗