TrafficLightRL leverages Reinforcement Learning (RL) to dynamically optimize traffic light control systems, reducing urban congestion and improving travel efficiency. This project showcases the power of machine learning in real-world infrastructure management, with a focus on scalability, safety, and adaptability.
- Kristian Diana - Project Lead
- Clara Wong - Project Member
- Ryan Li - Project Member
- Varun Pathak - Project Member
- Tridib Banik -- Project Member
- Smart Traffic Control: Adaptive traffic light decisions powered by RL agents.
- Real-World Simulations: Authentic intersection models using SUMO and OpenStreetMap.
- Custom Reward Functions: Tailored metrics to balance traffic flow and safety.
- University-Specific Demos: Interactive optimizations for Ontario campuses.
- Comprehensive Visualizations: Progress tracking with SUMO-GUI simulations.
- Reinforcement Learning: Stable-Baselines3 for agent training.
- Traffic Simulation: SUMO (Simulation of Urban MObility) for realistic simulations.
- Environment Design: OpenAI Gymnasium for RL integration.
- Mapping Tools: OpenStreetMap for real-world data.
Description: will probably include a short video demo of the MVP as well as a bunch of metrics (big numbers that will help make our project stand out!) Ex. Reduced wait times by 30%
After the completion of our MVP, we decided to expand upon the complexity of our project utilizing the OpenStreetMap toolbox! We applied TrafficLightRL to various University campuses across Canada to demonstrate how our project might have an impact on real students' lives. This served as our interactive demonstration component when presenting at CUCAI 2025!
- 🏙️ University of Toronto Read more about Varun's work here!
- 🏗️ University of Waterloo Read more about Clara's work here!
- 🛤️ Queen's University Learn more about Ryan's work here!
- 🚏 Western University Learn more about Tridib's Work here!
- 🏠 McMaster University Learn more about Kristian's work here!
Dive into our Design Document for an in-depth description of the project timeline from MVP to final product, and discussions of the design choices we made along the way.