How mibha/project-hero works
Project type and purpose
• Project HERO is a standalone AI-powered optimization system built in Python. It integrates a high-fidelity traffic simulator (SUMO) with real-world map data (OpenStreetMap) to train Multi-Agent Reinforcement Learning (MARL) agents. Its purpose is to minimize emergency response times via cooperative dispatch.
File structure
• Top-level directories:
• data/ – contains historical incident prediction datasets
• notebooks/ – exploratory Jupyter notebooks for analysis and prototyping
• simulation_files/ – SUMO network maps, route files, and configuration XMLs
• src/ – core Python modules implementing environment wrappers, agent code, training loops, and utility functions
• Configuration/build files:
• requirements.txt – Python dependencies
• .gitignore
• README.md
README insights
• Core architecture:
• Simulation engine (SUMO) ↔ Python via TraCI
• MARL training using Ray RLlib
• Data preprocessing and predictive hotspot modeling with Pandas/Scikit-learn
Detail
• Blocks with key technologies (e.g., “Ray RLlib”, “TraCI API”)
• Both offline (training) and online (inference/dispatch) flows