SpaceLayoutGym Environment

This repository contains an OpenAI Gym compatible environment for space layout design.

Installation

1. Clone this repository:

   git clone https://github.com/RezaKakooee/space_layout_gym.git
   
   

2. Create and activate a Conda environment:

   conda create -n slg python=3.9
   conda activate slg
   

3. Install the environment:

   cd space_layout_gym
   pip install -e gym-floorplan
   

4. Install required packages:

   pip install -r requirements.txt
   

5. Create a .env file in the root directory of the project and add the following variable:

   HOUSING_DESIGN_ROOT_DIR=path/to/space_layout_gym
   

   Replace path/to/space_layout_gym with the actual path to your space_layout_gym directory.

Usage

Here's a basic example of how to use the environment:

from gym_floorplan.envs.fenv_config import LaserWallConfig
from gym_floorplan.envs.master_env import SpaceLayoutGym

fenv_config = LaserWallConfig().get_config()
env = SpaceLayoutGym(fenv_config)

obs = env.reset()

done = False
while not done:
    action = env.action_space.sample()
    obs, reward, done, truncated, info = env.step(action)
    env.render()
env.close()

Training code

TODO

• We used various Deep RL libraries including RLlib, StableBaselines3, CleanRL and our custom built RL algorithms. However, we do not share the trainig code as we are still working on this research. Nonetheless, we provide initial codes examples showing how one can use RLlib and StableBaselines3 with SpaceLayoutGym environment.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Citation

@article{kakooee2024reimagining,
  title={Reimagining space layout design through deep reinforcement learning},
  author={Kakooee, Reza and Dillenburger, Benjamin},
  journal={Journal of Computational Design and Engineering},
  pages={qwae025},
  year={2024},
  publisher={Oxford University Press}
}