Files:
- agent.py : script to define agent and generate a policy
- design_obstacles.py : use PyGame to design goals and obstacles via point-and-click
- grid.py : define the grid-world and optionally, its rewards
- main.py : main driver script that shows an example
- plotting.py : generate plots for the reward and grid-world
If you use this grid-world setup, please consider citing:
@misc{puranic_dw,
author = "Aniruddh G. Puranic and Anand Balakrishnan",
title = "Reinforcement Learning Environment for Multiple Tasks",
year = 2020,
url = {https://github.com/CPS-VIDA/discrete-world}
}
This environment was used to generate plots for experiments in the paper Learning from Demonstrations using Signal Temporal Logic.