Inspiration
Traffic congestion wastes time, fuel, and increases COâ‚‚ emissions. We were inspired to create a solution that improves traffic flow by using the infrastructure already available in cities.
What it does
IRIS uses AI to analyze traffic from existing CCTV feeds and adjusts signal timings dynamically to reduce waiting time and emissions.
How we built it
We used YOLOv11x for vehicle detection and reinforcement learning to optimize signal timing. The system processes real time data and makes the best timing decisions for the intersection.
Challenges we ran into
Collecting reliable traffic data, handling low quality CCTV footage, and designing a reward function that balances efficiency and fairness were major challenges.
Accomplishments that we're proud of
A functional prototype that demonstrates automatic traffic optimization with measurable improvements in flow and a modular design that is scalable across intersections.
What we learned
We learned practical integration of RL with computer vision, efficient edge AI deployment, and how small timing improvements can significantly reduce emissions.
What's next for IRIS : Intelligent Roadway Infrastructure System
Deploying multi intersection control, adding emergency vehicle prioritization, and testing in collaboration with urban authorities for real world validation.
Log in or sign up for Devpost to join the conversation.