online-xiang-qi

An online Xiang Qi (Chinese Chess) game web app backend

Description

The project is a web application implements an AI using Monte Carlo Tree Search algorithm with a Convolutional Neural Network in a Chinese Chess Game.
The tree-based AI has a reasonable performance when the rollout times is larger than 50 and depth limit is larger than 100. However, the tree-based AI integrated with NN has worse performance than the tree-based AI without NN, which is possibly due to the inappropriate utility evaluation and the incorrect design of NN model.
The project contains three modules: backend server, frontend application, and NN module.

Get Started

Prerequisites

• Docker (version = 20.10.17)

Deployment

Backend

1. Clone the repository: $ git clone https://github.com/snail-inO/online-xiang-qi.git
2. Go to the project code directory: $ cd online-xiang-qi/code
3. Start Docker client
4. Run containers: docker-compose up (need to run after building Docker image of frontend, Frontend step4)

Frontend

1. Clone the repository: $ git clone https://github.com/snail-inO/online-xiang-qi-frontend.git
2. Go to the project directory: $ cd online-xiang-qi-frontend
3. Start Docker client
4. Build docker image: $ docker build -t front .

Usage

1. Access http://localhost:3000 on a browser
2. Enter a user name
3. Select game modes and problem size (opponet mode cannot set to user mode if your mode isn't user mode)
4. Red player moves first
5. Perform a move
6. (User mode) Click on the piece of your color to select the piece (click one more time to unselect) and click on a valid position on the board to move
7. (AI mode) Click Next button to perform a move (Do not click the button if opponent's move is not completed)
8. Wait until opponent's move is completed, then repeat step 5
9. If you do not perform an action longer than 60 seconds, you will lose the game

Licence

This project is licensed under the GPLv2 License.