Automated Simulation of TRAnsportation networks for SUMO
ASTRA turns real-world traffic data and natural-language tasks into SUMO simulations.
Installation · Documentation · Examples · MIT License
| Build SUMO networks from OpenStreetMap, vehicle trajectory data, and road construction maps. | Reconstruct traffic demand and calibrate simulations using real sensor measurements. | Run further SUMO experiments from natural-language instructions. |
codex plugin marketplace add Tarard/ASTRA-SUMO --ref main
codex plugin add astra-sumo@astra-sumoThen ask Codex, for example:
Use ASTRA to build a SUMO network from this OSM area.
Check connectivity, traffic signals, and routeability.
ASTRA requires Python 3.11+ and Eclipse SUMO.
ASTRA continues Torii in this new repository. See the migration guide for installation changes and compatibility.
ASTRA is being used to reconstruct and validate a real traffic corridor in central Hamburg.
ASTRA V1 corridor over aerial imagery, with lane connections and virtual sensors.
ASTRA combines official traffic data, aerial imagery, and SUMO network reconstruction in one workflow.
LSA118 reconstruction: official MAP endpoints and headings, reconstructed curves, and lane connections before and after cleaning.
The current Hamburg calibration matches the aggregate detector count exactly, with 0.15 vehicles MAE per 15-minute bin.
Architecture · Installation · Documentation · Examples
ASTRA-SUMO is licensed under the MIT License.
ASTRA 1.3.1 is archived on Zenodo. The project DOI covers all ASTRA versions.
Earlier Torii releases remain in their original Zenodo archive.