Inspiration

  • Hide and Seek AI, developed by OpenAI
  • Inspired us to make a simple game and develop an intelligent agent to play the game

What it does

  • The agent can learn the best strategy to win the game against its opponent
  • The game supports 2-player mode
  • Intelligent agents training was unsuccessful, the agents did not learn the optimum strategy for the game

How we built it

  • The program is written in Python
  • We used Tkinter to create a GUI and the rules for the game
  • We used PyTorch to develop the neural network model, and model training

Challenges we ran into

  • Training of the network failed to converge
  • Using Tkinter as a GYM training environment

Accomplishments that we're proud of

  • Managed to finish the tasks assign, which are making the game, building the intelligent agent and training the model
  • Created a fun game and got the model trained

What we learned

  • Using Tkinter as a GUI tool for simple game making
  • Reinforcement learning

What's next for TileGridAI

  • Build an intelligent agent that can successfully learn the game
  • Make the game more interesting and challenging

Built With

Share this project:

Updates