Worthify Community

This project is a local decision-support prototype for studying coordinated traffic signals on a real Korean road map. It combines OpenStreetMap (OSM) roads and junctions, Korean government intersection reference data, SUMO traffic simulation, and a joint reinforcement-learning controller. The Streamlit dashboard guides a user through choosing a study area and traffic scenario, running a simulation, and comparing the proposed controller with a SUMO fixed-time reference. The city-scale controller's primary learning objective is network throughput, measured as completed trips per simulated hour. Queue length, waiting time, and trip time are additional checks. Generated routes and scenario demand are synthetic or assumption-based unless explicitly identified as imported observations. This is a research prototype: it does not control roadside equipment, submit timing changes to a government system, or establish real-world performance. The current default is a Seoul test area around Jongno-gu. The city selector is driven by config/korean_regions.json; availability of government records can vary by city. OSM roads, junction candidates, and potential traffic generators are displayed on the map. Generator importance is a map-category estimate, not a measured traffic count.