Open Issues Need Help
View All on GitHubAI Summary: Implement TensorBoard or Weights & Biases logging for the OpenTrafficAI project's training script. This involves adding a command-line argument to select the logging method and recording episode return, average queue length, and exploration rate. Finally, update the README with a visual example of the logging.
OpenTrafficAI is an open‑source Python toolkit for adaptive traffic‑signal control with Reinforcement Learning.
AI Summary: Create a MkDocs-based documentation site for the OpenTrafficAI project, including auto-generated API documentation using mkdocstrings, and pages for quick-start, API reference, and contributing. The site should be deployed to GitHub Pages upon pushes to the main branch.
OpenTrafficAI is an open‑source Python toolkit for adaptive traffic‑signal control with Reinforcement Learning.
AI Summary: Enhance the `visualize.py` script within the OpenTrafficAI project to generate plots of episode statistics from a CSV file. These plots should include total reward per episode, queue length per intersection over time, and a phase (green/red) timeline. Only standard matplotlib is allowed.
OpenTrafficAI is an open‑source Python toolkit for adaptive traffic‑signal control with Reinforcement Learning.
AI Summary: Implement a Proximal Policy Optimization (PPO) agent for adaptive traffic signal control within the OpenTrafficAI Python toolkit. This involves coding the PPO algorithm (including clipping, GAE, and entropy bonus), writing unit tests, and integrating it into the existing training script with a command-line flag. The agent's performance will be evaluated based on achieving a negative total queue below -500 within 10,000 training episodes.
OpenTrafficAI is an open‑source Python toolkit for adaptive traffic‑signal control with Reinforcement Learning.