Deep-Flow-Optimizer: Adaptive Signal Control with Deep RL

Deep-Flow-Optimizer is an intelligent traffic management system built to solve complex urban congestion and bridge spillback. Utilizing Deep Q-Learning (DQN) and the SUMO simulator, this project demonstrates how AI agents can autonomously manage signal phases based on real-time lane-level data.

🛠 Project Overview

Urban intersections often suffer from downstream "spillback" that causes total network gridlock. This project benchmarks three control strategies:

1. Fixed-Time: Traditional rigid signal cycles.
2. Tabular Q-Learning: An RL agent using discrete state-mapping.
3. Deep Q-Network (DQN): A neural-network-based agent for high-dimensional state spaces.

📈 Final Performance Benchmarks

Metric	Fixed-Time	Q-Learning	Deep Q-Network
Total Throughput	154 veh	157 veh	165 veh
Network Delay	381,959s	283,253s	472,157s
CO2 Impact	Baseline	-1.4% (Saving)	+5.3% (Learning Phase)

📊 Visual Results

Traffic Load & Queue Management

System Reward (Learning Curve)

🚀 Deployment

1. Install SUMO.
2. Install Python dependencies: pip install -r requirements.txt.
3. Run the DQL agent: python DQL_Agent.py.

📝 Conclusion

While the DQN agent maximized vehicle throughput (165 arrived), the Tabular Q-Learning agent achieved the highest overall efficiency with a 25.8% reduction in delay. This suggests that for single-intersection optimization, lower-dimensional state mappings converge more effectively within short training windows.