OpenTrafficAI

OpenTrafficAI is an open‑source Python toolkit for adaptive traffic‑signal control with Reinforcement Learning.

• Multi‑intersection Gymnasium environment simulating traffic queues.
• Reference agents implemented in PyTorch (DQN today, PPO stub for extension).
• Training / evaluation scripts, simple visualisation, and unit‑tests.
• Docker & GitHub CI for easy reproducibility.

Quick start

git clone https://github.com/andbalashov/OpenTrafficAI.git
cd OpenTrafficAI
python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
python -m traffic_ai.scripts.train --episodes 1000 --n 4 --render